Search

Research paper recommenders emerged over the last decade to ease finding publications relating to researchers’ area of interest. The challenge was not just to provide researchers with very rich publications at any time, any place and in any form but to also offer the right publication to the right researcher in the right way. Several approaches exist in handling paper recommender systems. However, these approaches assumed the availability of the whole contents of the recommending papers to be freely accessible, which is not always true due to factors such as copyright restrictions. This paper presents a collaborative approach for research paper recommender system. By leveraging the advantages of collaborative filtering approach, we utilize the publicly available contextual metadata to infer the hidden associations that exist between research papers in order to personalize recommendations. The novelty of our proposed approach is that it provides personalized recommendations regardless of the research field and regardless of the user’s expertise. Using a publicly available dataset, our proposed approach has recorded a significant improvement over other baseline methods in measuring both the overall performance and the ability to return relevant and useful publications at the top of the recommendation list. [ABSTRACT FROM AUTHOR]

The retrieval of important information from a dataset requires applying a special data mining technique known as data clustering (DC). DC classifies similar objects into a groups of similar characteristics. Clustering involves grouping the data around k-cluster centres that typically are selected randomly. Recently, the issues behind DC have called for a search for an alternative solution. Recently, a nature-based optimization algorithm named Black Hole Algorithm (BHA) was developed to address the several well-known optimization problems. The BHA is a metaheuristic (population-based) that mimics the event around the natural phenomena of black holes, whereby an individual star represents the potential solutions revolving around the solution space. The original BHA algorithm showed better performance compared to other algorithms when applied to a benchmark dataset, despite its poor exploration capability. Hence, this paper presents a multi-population version of BHA as a generalization of the BHA called MBHA wherein the performance of the algorithm is not dependent on the best-found solution but a set of generated best solutions. The method formulated was subjected to testing using a set of nine widespread and popular benchmark test functions. The ensuing experimental outcomes indicated the highly precise results generated by the method compared to BHA and comparable algorithms in the study, as well as excellent robustness. Furthermore, the proposed MBHA achieved a high rate of convergence on six real datasets (collected from the UCL machine learning lab), making it suitable for DC problems. Lastly, the evaluations conclusively indicated the appropriateness of the proposed algorithm to resolve DC issues. [ABSTRACT FROM AUTHOR]

In this paper, we create EMIR, the first-ever Music Information Retrieval dataset for Ethiopian music. EMIR is freely available for research purposes and contains 600 sample recordings of Orthodox Tewahedo chants, traditional Azmari songs and contemporary Ethiopian secular music. Each sample is classified by five expert judges into one of four well-known Ethiopian Kiñits, Tizita, Bati, Ambassel and Anchihoye. Each Kiñit uses its own pentatonic scale and also has its own stylistic characteristics. Thus, Kiñit classification needs to combine scale identification with genre recognition. After describing the dataset, we present the Ethio Kiñits Model (EKM), based on VGG, for classifying the EMIR clips. In Experiment 1, we investigated whether Filterbank, Mel-spectrogram, Chroma, or Mel-frequency Cepstral coefficient (MFCC) features work best for Kiñit classification using EKM. MFCC was found to be superior and was therefore adopted for Experiment 2, where the performance of EKM models using MFCC was compared using three different audio sample lengths. 3s length gave the best results. In Experiment 3, EKM and four existing models were compared on the EMIR dataset: AlexNet, ResNet50, VGG16 and LSTM. EKM was found to have the best accuracy (95.00%) as well as the fastest training time. However, the performance of VGG16 (93.00%) was found not to be significantly worse (P < 0.01). We hope this work will encourage others to explore Ethiopian music and to experiment with other models for Kiñit classification. [ABSTRACT FROM AUTHOR]

The world of science is growing at an unprecedented speed with more and more scholarly papers produced each year. The scientific landscape is constantly changing as research specialties evolve, merge or become obsolete. It is difficult for researchers, research managers and the public alike to keep abreast with these changes and maintain a true and fair overview of the world of science. Such an overview is necessary to stimulate scientific progress, to maintain flexible and responsive research organizations, and to secure collaboration and knowledge exchange between different research specialties and the wider community. Although science mapping is applied to a wide range of scientific areas, examples of their practical use are sparse. This paper demonstrates how to use a topical, scientific reference maps to understand and navigate in dynamic research landscapes and how to utilize science maps to facilitate strategic thinking. In this study, the research domain of biology at Aarhus University serves as an example. All scientific papers authored by the current, permanent staff were extracted (6,830 in total). These papers were used to create a semantic cognitive map of the research field using a co-word analysis based on keywords and keyword phrases. A workflow was written in Python for easy and fast retrieval of information for topic maps (including tokens from keywords section and title) to generate intelligible research maps, and to visualize the distribution of topics (keywords), papers, journal categories, individual researchers and research groups on any scale. The resulting projections revealed new insights into the structure of the research community and made it possible to compare researchers or research groups to describe differences and similarities, to find scientific overlaps or gaps, and to understand how they relate and connect. Science mapping can be used for intended (top-down) as well as emergent (bottom-up) strategy development. The paper concludes that science maps provide alternative views of the intricate structures of science to supplement traditional bibliometric information. These insights may help strengthen strategic thinking and boost creativity and thus contribute to the progress of science. [ABSTRACT FROM AUTHOR]

Semantic Textual Similarity (STS) is the task of identifying the semantic correlation between two sentences of the same or different languages. STS is an important task in natural language processing because it has many applications in different domains such as information retrieval, machine translation, plagiarism detection, document categorization, semantic search, and conversational systems. The availability of STS training and evaluation data resources for some languages such as English has led to good performance systems that achieve above 80% correlation with human judgment. Unfortunately, such required STS data resources are not available for many languages like Arabic. To overcome this challenge, this paper proposes three different approaches to generate effective STS Arabic models. The first one is based on evaluating the use of automatic machine translation for English STS data to Arabic to be used in fine-tuning. The second approach is based on the interleaving of Arabic models with English data resources. The third approach is based on fine-tuning the knowledge distillation-based models to boost their performance in Arabic using a proposed translated dataset. With very limited resources consisting of just a few hundred Arabic STS sentence pairs, we managed to achieve a score of 81% correlation, evaluated using the standard STS 2017 Arabic evaluation set. Also, we managed to extend the Arabic models to process two local dialects, Egyptian (EG) and Saudi Arabian (SA), with a correlation score of 77.5% for EG dialect and 76% for the SA dialect evaluated using dialectal conversion from the same standard STS 2017 Arabic set. [ABSTRACT FROM AUTHOR]

The Information Retrieval user experience has remained largely unchanged since its inception for computers and mobile devices alike. However, recent developments in Virtual Reality hardware (pioneered by Oculus Rift in 2013) could introduce a new environment for Information Retrieval. This paper reports the results of a Scoping Literature Review (PRISMA-ScR) by rigorously examining the entire body of relevant literature with reproducible methods. The following research questions are answered: "What prototypes and concepts of Virtual Reality Information Retrieval systems with current generation hardware exist?", "How are user interaction and especially user input realised in these systems?", "What Retrieval features are used in these systems?", "How are search results displayed in these systems?" and "Can these VR IR systems compare to traditional (non-VR) IR systems?". After querying Google Scholar, Scopus and Web of Science, 1042 documents were reviewed in depth. Key features and attributes of the systems are summarised and discussed. Sketches of the user interfaces are included as well. The 30 documents that were relevant to the research questions include 16 distinct systems or theories. They discuss and utilise several user input technologies, ranging from controllers, voice input or hand tracking. Although conventional retrieval features are less common, systems enable retrieval of literature, 3D objects, images, books and texts and arrange them in a virtual space (e.g. as grids, arcs or maps). Finally, many of these systems were compared to conventional counterparts through user evaluation (n = 10). Most found user task times to be shorter or equal (n = 5, n = 3). In the seven papers that measured user performance (rate of correct solutions), three reported better performance (one equal). Notably, users always were more satisfied with the Virtual Reality systems compared to conventional ones. Possible limitations of these evaluations are demographic selection and the quality of baseline systems (control). [ABSTRACT FROM AUTHOR]

Cryptographic cloud storage is used to make optimal use of the cloud storage infrastructure to outsource sensitive and mission-critical data. The continuous growth of encrypted data outsourced to cloud storage requires continuous updating. Attacks like file-injection are reported to compromise confidentiality of the user as a consequence of information leakage during update. It is required that dynamic schemes provide forward privacy guarantees. Updates should not leak information to the untrusted server regarding the previously issued queries. Therefore, the challenge is to design an efficient searchable encryption scheme with dynamic updates and forward privacy guarantees. In this paper, a novel private multi-linked dynamic index for encrypted document retrieval namely Pindex is proposed. The multi-linked dynamic index is constructed using probabilistic homomorphic encryption mechanism and secret orthogonal vectors. Full security proofs for correctness and forward privacy in the random oracle model is provided. Experiments on real world Enron dataset demonstrates that our construction is practical and efficient. The security and performance analysis of Pindex shows that the dynamic multi-linked index guarantees forward privacy without significant loss of efficiency. [ABSTRACT FROM AUTHOR]

Purpose: The credibility and the reliability of Internet webpages to seek medication-related information is questionable. The main objective of the current study was to evaluate perception and experience of pharmacists with the use of Internet-based medication information by their patients. Methods: This is a cross-sectional descriptive study that was conducted to evaluate perception and experience of pharmacists with the use of Internet-based medication information by their patients. During the study period, 200 pharmacists were approached to participate in the study using a paper-based survey to assess their perceptions and current experience with the use of Internet-based medication information by their patients. Data were analyzed using descriptive statistics (mean/standard deviation for continuous variables, and frequency/percentages for qualitative variables). Also, simple linear regression was utilized to screen factors affecting pharmacists' perception scores of the use of Internet-based medication information. Results: Among 161 recruited pharmacists, the majority (n = 129, 80.1%) reported receiving inquiries from patients about Internet-based medication information within the last year. Among them, only 22.6% (n = 29) of pharmacists believed that Internet-based medication information is somewhat or very accurate. Unfortunately, only 24.2% (n = 31) of them stated that they always had enough time for their patient to discuss their Internet-based medication information. Regarding pharmacists' perception of the use of Internet-based medication information by their patients, more than half of the pharmacists (>50%) believe that Internet-based medication information could increase the patient's role in taking responsibility. On the other hand, 54.7% (n = 88) of the pharmacists believed that Internet-based medication information would contribute to rising the healthcare cost by obtaining unnecessary medications by patients. Finally, pharmacists' educational level was found to significantly affect their perception scores toward patient use of Internet-based medication information where those with higher educational level showed lower perception score (r = -0.200, P-value = 0.011). Conclusion: Although pharmacists felt that usage of Internet-based data by patients is beneficial, they also have believed that it has a negative impact in terms of rising the healthcare cost, and it promotes unnecessary fear or concern about medications. We suggest that pharmacists be trained on principles of critical appraisal to become professional in retrieval information on the Internet that might improve their delivery of healthcare information and their recommendations to patients. [ABSTRACT FROM AUTHOR]

Due to the rapid growth of information available about individual patients, most physicians suffer from information overload and inefficiencies when they review patient information in health information technology systems. In this paper, we present a novel hybrid dynamic and multi-collaborative filtering method to improve information retrieval from electronic health records. This method recommends relevant information from electronic health records to physicians during patient visits. It models information search dynamics using a Markov model. It also leverages the key idea of collaborative filtering, originating from Recommender Systems, for prioritizing information based on various similarities among physicians, patients and information items. We tested this new method using electronic health record data from the Indiana Network for Patient Care, a large, inter-organizational clinical data repository maintained by the Indiana Health Information Exchange. Our experimental results demonstrated that, for top-5 recommendations, our method was able to correctly predict the information in which physicians were interested in 46.7% of all test cases. For top-1 recommendations, the corresponding figure was 24.7%. In addition, the new method was 22.3% better than the conventional Markov model for top-1 recommendations. [ABSTRACT FROM AUTHOR]

This study compared the results of data collected from a longitudinal query analysis of the MEDLINE database hosted on multiple platforms that include PubMed, EBSCOHost, Ovid, ProQuest, and Web of Science. The goal was to identify variations among the search results on the platforms after controlling for search query syntax. We devised twenty-nine cases of search queries comprised of five semantically equivalent queries per case to search against the five MEDLINE database platforms. We ran our queries monthly for a year and collected search result count data to observe changes. We found that search results varied considerably depending on MEDLINE platform. Reasons for variations were due to trends in scholarly publication such as publishing individual papers online first versus complete issues. Some other reasons were metadata differences in bibliographic records; differences in the levels of specificity of search fields provided by the platforms and large fluctuations in monthly search results based on the same query. Database integrity and currency issues were observed as each platform updated its MEDLINE data throughout the year. Specific biomedical bibliographic databases are used to inform clinical decision-making, create systematic reviews, and construct knowledge bases for clinical decision support systems. They serve as essential information retrieval and discovery tools to help identify and collect research data and are used in a broad range of fields and as the basis of multiple research designs. This study should help clinicians, researchers, librarians, informationists, and others understand how these platforms differ and inform future work in their standardization. [ABSTRACT FROM AUTHOR]

Recently. recommender systems have become a very crucial application in the online market and e-commerce as users are often astounded by choices and preferences and they need help finding what the best they are looking for. Recommender systems have proven to overcome information overload issues in the retrieval of information, but still suffer from persistent problems related to cold-start and data sparsity. On the flip side, sentiment analysis technique has been known in translating text and expressing user preferences. It is often used to help online businesses to observe customers' feedbacks on their products as well as try to understand customer needs and preferences. However, the current solution for embedding traditional sentiment analysis in recommender solutions seems to have limitations when involving multiple domains. Therefore, an issue called domain sensitivity should be addressed. In this paper, a sentiment-based model with contextual information for recommender system was proposed. A novel solution for domain sensitivity was proposed by applying a contextual information sentiment-based model for recommender systems. In evaluating the contributions of contextual information in sentiment-based recommendations, experiments were divided into standard rating model, standard sentiment model and contextual information model. Results showed that the proposed contextual information sentiment-based model illustrates better performance as compared to the traditional collaborative filtering approach. [ABSTRACT FROM AUTHOR]

Set containment operations form an important tool in various fields such as information retrieval, AI systems, object-relational databases, and Internet applications. In the paper, a set-trie data structure for storing sets is considered, along with the efficient algorithms for the corresponding set containment operations. We present the mathematical and empirical study of the set-trie. In the mathematical study, the relevant upper-bounds on the efficiency of its expected performance are established by utilizing a natural probabilistic model. In the empirical study, we give insight into how different distributions of input data impact the efficiency of set-trie. Using the correct parameters for those randomly generated datasets, we expose the key sources of the input sensitivity of set-trie. Finally, the empirical comparison of set-trie with the inverted index is based on the real-world datasets containing sets of low cardinality. The comparison shows that the running time of set-trie consistently outperforms the inverted index by orders of magnitude. [ABSTRACT FROM AUTHOR]

Biomedical question answering (QA) represents a growing concern among industry and academia due to the crucial impact of biomedical information. When mapping and ranking candidate snippet answers within relevant literature, current QA systems typically refer to information retrieval (IR) techniques: specifically, query processing approaches and ranking models. However, these IR-based approaches are insufficient to consider both syntactic and semantic relatedness and thus cannot formulate accurate natural language answers. Recently, deep learning approaches have become well-known for learning optimal semantic feature representations in natural language processing tasks. In this paper, we present a deep ranking recursive autoencoders (rankingRAE) architecture for ranking question-candidate snippet answer pairs (Q-S) to obtain the most relevant candidate answers for biomedical questions extracted from the potentially relevant documents. In particular, we convert the task of ranking candidate answers to several simultaneous binary classification tasks for determining whether a question and a candidate answer are relevant. The compositional words and their random initialized vectors of concatenated Q-S pairs are fed into recursive autoencoders to learn the optimal semantic representations in an unsupervised way, and their semantic relatedness is classified through supervised learning. Unlike several existing methods to directly choose the top-K candidates with highest probabilities, we take the influence of different ranking results into consideration. Consequently, we define a listwise "ranking error" for loss function computation to penalize inappropriate answer ranking for each question and to eliminate their influence. The proposed architecture is evaluated with respect to the BioASQ 2013-2018 Six-year Biomedical Question Answering benchmarks. Compared with classical IR models, other deep representation models, as well as some state-of-the-art systems for these tasks, the experimental results demonstrate the robustness and effectiveness of rankingRAE. [ABSTRACT FROM AUTHOR]

With the convergence of various broadcasting systems, the amount of content available in mobile terminals including IPTV has significantly increased. In this paper, we propose a system that enables users to schedule programs considering both section video clips and full video clips based on the user detection method with similar preference. And, since the system constituting the contents can be classified according to the program, the proposed method can store a program desired by the user, and thus create and schedule a kind of individual channel. Experimental results show that the proposed method has a higher prediction accuracy; this is accomplished by comparing existing channel recommendation methods with the program recommendation methods proposed in this paper. [ABSTRACT FROM AUTHOR]

Several network features and information retrieval methods have been proposed to elucidate the structure of citation networks and to detect important nodes. However, it is difficult to retrieve information related to trends in an academic field and to detect cutting-edge areas from the citation network. In this paper, we propose a novel framework that detects the trend as the growth direction of a citation network using network representation learning(NRL). We presume that the linear growth of citation network in latent space obtained by NRL is the result of the iterative edge additional process of a citation network. On APS datasets and papers of some domains of the Web of Science, we confirm the existence of trends by observing that an academic field grows in a specific direction linearly in latent space. Next, we calculate each node’s degree of trend-following as an indicator called the intrinsic publication year (IPY). As a result, there is a correlation between the indicator and the number of future citations. Furthermore, a word frequently used in the abstracts of cutting-edge papers (high-IPY paper) is likely to be used often in future publications. These results confirm the validity of the detected trend for predicting citation network growth. [ABSTRACT FROM AUTHOR]

A number of publishers and funders, including PLOS, have recently adopted policies requiring researchers to share the data underlying their results and publications. Such policies help increase the reproducibility of the published literature, as well as make a larger body of data available for reuse and re-analysis. In this study, we evaluate the extent to which authors have complied with this policy by analyzing Data Availability Statements from 47,593 papers published in PLOS ONE between March 2014 (when the policy went into effect) and May 2016. Our analysis shows that compliance with the policy has increased, with a significant decline over time in papers that did not include a Data Availability Statement. However, only about 20% of statements indicate that data are deposited in a repository, which the PLOS policy states is the preferred method. More commonly, authors state that their data are in the paper itself or in the supplemental information, though it is unclear whether these data meet the level of sharing required in the PLOS policy. These findings suggest that additional review of Data Availability Statements or more stringent policies may be needed to increase data sharing. [ABSTRACT FROM AUTHOR]

Background: \Anticoagulation is used for stroke prophylaxis in non-valvular atrial fibrillation, amongst other by use of the vitamin K antagonist, warfarin. Quality in warfarin therapy is often summarized by the time patients spend within the therapeutic range (percent time in therapeutic range, TTR). The correlation between TTR and the occurrence of complications during warfarin therapy has been established, but the influence of patient characteristics in that respect remains undetermined. The objective of the present papers was to examine the association between mean TTR and complication rates with adjustment for differences in relevant patient cohort characteristics. Methods: A systematic literature search was conducted in MEDLINE and Embase (2005–2015) to identify eligible studies reporting on use of warfarin therapy by patients with non-valvular atrial fibrillation and the occurrence of hemorrhage and thromboembolism. Both randomized controlled trials and observational cohort studies were included. The association between the reported mean TTR and major bleeding and stroke/systemic embolism was analyzed by random-effects meta-regression with and without adjustment for relevant clinical cohort characteristics. In the adjusted meta-regressions, the impact of mean TTR on the occurrence of hemorrhage was adjusted for the mean age and the proportion of populations with prior stroke or transient ischemic attack. In the adjusted analyses on thromboembolism, the proportion of females was, furthermore, included. Results: Of 2169 papers, 35 papers met pre-specified inclusion criteria, holding relevant information on 31 patient cohorts. In univariable meta-regression, increasing mean TTR was significantly associated with a decreased rate of both major bleeding and stroke/systemic embolism. However, after adjustment mean TTR was no longer significantly associated with stroke/systemic embolism. The proportion of residual variance composed by between-study heterogeneity was substantial for all analyses. Conclusions: Although higher mean TTR in warfarin therapy was associated with lower complication rates in atrial fibrillation, the strength of the association was decreased when adjusting for differences in relevant clinical characteristics of the patient cohorts. This study suggests that mainly the safety of warfarin therapy increases with higher mean TTR, whereas effectiveness appears not to be substantially improved. Due to the limitations immanent in the meta-regression methods, the results of the present study should be interpreted with caution. Further research on the association between the quality of warfarin therapy and risk of complications is warranted with adjustment for clinically relevant characteristics. [ABSTRACT FROM AUTHOR]

The paper presents a new compositional hierarchical model for robust music transcription. Its main features are unsupervised learning of a hierarchical representation of input data, transparency, which enables insights into the learned representation, as well as robustness and speed which make it suitable for real-world and real-time use. The model consists of multiple layers, each composed of a number of parts. The hierarchical nature of the model corresponds well to hierarchical structures in music. The parts in lower layers correspond to low-level concepts (e.g. tone partials), while the parts in higher layers combine lower-level representations into more complex concepts (tones, chords). The layers are learned in an unsupervised manner from music signals. Parts in each layer are compositions of parts from previous layers based on statistical co-occurrences as the driving force of the learning process. In the paper, we present the model’s structure and compare it to other hierarchical approaches in the field of music information retrieval. We evaluate the model’s performance for the multiple fundamental frequency estimation. Finally, we elaborate on extensions of the model towards other music information retrieval tasks. [ABSTRACT FROM AUTHOR]

This paper proposes a method to predict fluctuations in the prices of cryptocurrencies, which are increasingly used for online transactions worldwide. Little research has been conducted on predicting fluctuations in the price and number of transactions of a variety of cryptocurrencies. Moreover, the few methods proposed to predict fluctuation in currency prices are inefficient because they fail to take into account the differences in attributes between real currencies and cryptocurrencies. This paper analyzes user comments in online cryptocurrency communities to predict fluctuations in the prices of cryptocurrencies and the number of transactions. By focusing on three cryptocurrencies, each with a large market size and user base, this paper attempts to predict such fluctuations by using a simple and efficient method. [ABSTRACT FROM AUTHOR]

The reuse of business processes (BPs) requires similarities between them to be suitably identified. Various approaches have been introduced to address this problem, but many of them feature a high computational cost and a low level of automation. This paper presents a clustering algorithm that groups business processes retrieved from a multimodal search system (based on textual and structural information). The algorithm is based on Incremental Covering Arrays (ICAs) with different alphabets to determine the possible number of groups to be created for each row of the ICA. The proposed algorithm also incorporates Balanced Bayesian Information Criterion to determine the optimal number of groups and the best solution for each query. Experimental evaluation shows that the use of ICAs with strength four (4) and different alphabets reduces the number of solutions needed to be evaluated and optimizes the number of clusters. The proposed algorithm outperforms other algorithms in various measures (precision, recall, and F-measure) by between 12% and 88%. Friedman and Wilcoxon non-parametric tests gave a 90–95% significance level to the obtained results. Better options of repository search for BPs help companies to reuse them. By thus reusing BPs, managers and analysts can more easily get to know the evolution and trajectory of the company processes, a situation that could be expected to lead to improved managerial and commercial decision making. [ABSTRACT FROM AUTHOR]

This paper focuses on fine-grained image retrieval based on sketches. Sketches capture detailed information, but their highly abstract nature makes visual comparisons with images more difficult. In spite of the fact that the existing models take into account the fine-grained details, they can not accurately highlight the distinctive local features and ignore the correlation between features. To solve this problem, we design a gradually focused bilinear attention model to extract detailed information more effectively. Specifically, the attention model is to accurately focus on representative local positions, and then use the weighted bilinear coding to find more discriminative feature representations. Finally, the global triplet loss function is used to avoid oversampling or undersampling. The experimental results show that the proposed method outperforms the state-of-the-art sketch-based image retrieval methods. [ABSTRACT FROM AUTHOR]

The accessing and processing of textual information (i.e. the storing and querying of a set of strings) is especially important for many current applications (e.g. information retrieval and social networks), especially when working in the fields of Big Data or IoT, which require the handling of very large string dictionaries. Typical data structures for textual indexing are Hash Tables and some variants of Tries such as the Double Trie (DT). In this paper, we propose an extension of the DT that we have called MergedTrie. It improves the DT compression by merging both Tries into a single and by segmenting the indexed term into two fixed length parts in order to balance the new Trie. Thus, a higher overlapping of both prefixes and suffixes is obtained. Moreover, we propose a new implementation of Tries that achieves better compression rates than the Double-Array representation usually chosen for implementing Tries. Our proposal also overcomes the limitation of static implementations that does not allow insertions and updates in their compact representations. Finally, our MergedTrie implementation experimentally improves the efficiency of the Hash Tables, the DTs, the Double-Array, the Crit-bit, the Directed Acyclic Word Graphs (DAWG), and the Acyclic Deterministic Finite Automata (ADFA) data structures, requiring less space than the original text to be indexed. [ABSTRACT FROM AUTHOR]

More than ever, technical inventions are the symbol of our society’s advance. Patents guarantee their creators protection against infringement. For an invention being patentable, its novelty and inventiveness have to be assessed. Therefore, a search for published work that describes similar inventions to a given patent application needs to be performed. Currently, this so-called search for prior art is executed with semi-automatically composed keyword queries, which is not only time consuming, but also prone to errors. In particular, errors may systematically arise by the fact that different keywords for the same technical concepts may exist across disciplines. In this paper, a novel approach is proposed, where the full text of a given patent application is compared to existing patents using machine learning and natural language processing techniques to automatically detect inventions that are similar to the one described in the submitted document. Various state-of-the-art approaches for feature extraction and document comparison are evaluated. In addition to that, the quality of the current search process is assessed based on ratings of a domain expert. The evaluation results show that our automated approach, besides accelerating the search process, also improves the search results for prior art with respect to their quality. [ABSTRACT FROM AUTHOR]

Objective: Recognizing what physicians know and do not know about a particular disease is one of the keys to designing clinical decision support systems, since these systems can fulfill complementary role by recognizing this boundary. To our knowledge, however, no study has attempted to quantify how many diseases physicians actually know and thus the boundary is unclear. This study explores a method to solve this problem by investigating whether the vocabulary assessment techniques developed in the linguistics field can be applied to assess physicians’ knowledge. Methods: The test design required us to pay special attention to disease knowledge assessment. First, to avoid imposing unnecessary burdens on the physicians, we chose a self-assessment questionnaire that was straightforward to fill out. Second, to prevent overestimation, we used a “pseudo-word” approach: fictitious diseases were included in the questionnaire, and positive responses to them were penalized. Third, we used paper-based tests, rather than computer-based ones, to further prevent participants from cheating by using a search engine. Fourth, we selectively used borderline diseases, i.e., diseases that physicians might or might not know about, rather than well-known or little-known diseases, in the questionnaire. Results: We collected 102 valid answers from 109 physicians who attended the seminars we conducted. On the basis of these answers, we estimated that the average physician knew of 2008 diseases (95% confidence interval: (1939, 2071)). This preliminary estimation agrees with the guideline for the national license examination in Japan, suggesting that this vocabulary assessment was able to evaluate physicians’ knowledge. The survey included physicians with various backgrounds, but there were no significant differences between subgroups. Other implication for researches on clinical decision support and limitation of the sampling method adopted in this study are also discussed, toward more rigorous estimation in future surveys. [ABSTRACT FROM AUTHOR]

The financial risk not only affects the development of the company itself, but also affects the economic development of the whole society; therefore, the financial risk assessment of company is an important part. At present, numerous methods of financial risk assessment have been researched by scholars. However, most of the extant methods neither integrated fuzzy sets with quantitative analysis, nor took into account the historical data of the past few years. To settle these defects, this paper proposes a novel financial risk assessment model for companies based on heterogeneous multiple-criteria decision-making (MCDM) and historical data. Subjective and objective indexes are comprehensively taken into consideration in the financial risk assessment index system of the model, which combines fuzzy theory with quantitative data analysis. Moreover, the assessment information obtained from historical financial information of company, credit rating agency and decision makers, including crisp numbers, triangular fuzzy numbers and neutrosophic numbers. Furthermore, the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method is used to determine the ranking order of companies according to their financial risk. Finally, an empirical study of financial risk assessment for companies is conducted, and the results of comparative analysis and sensitivity analysis suggest that the proposed model can effectively and reliably obtain the company with the lowest financial risk. [ABSTRACT FROM AUTHOR]

The notable increase in the size and dimensions of data have presented challenges for data storage and retrieval. The Bloom filter and its generations, due to efficient space overheads and constant query delays, have been broadly applied to querying memberships of a big data set. However, the Bloom filter and most of the variants regard each element as a 1-dimensional string and adopt multiple different string hashes to project the data. The interesting problem is when the inputs are numerical vectors with high dimensions, it remains unknown whether they can be projected into the Bloom filter in their original format. Furthermore, we investigate whether the projection is random and uniform. To address these problems, this paper presents a new uniform Prime-HD-BKDERhash family and a new Bloom filter (P-HDBF) to retrieve the membership of a big data set with the numerical high dimensions. Since the randomness and uniformity of data mapping determines the performance of the Bloom filter, to verify these properties, we first introduce information entropy. Our theoretical and experimental results show that the P-HDBF can randomly and uniformly map the data in their native formats. Moreover, the P-HDBF provides an efficient solution alternative to implement membership search with space-time overheads. This advantage may be suitable for engineering applications that are resource-constrained or identification of the nuances of the graphics and images. [ABSTRACT FROM AUTHOR]

The shift of the Noncommunicable Diseases (NCDs) epidemic, including cardiovascular disease, from developed to Low and Middle Income Countries (LMIC), creates new challenges in contexts where there is poor information on healthcare costs. Clearly this information is essential for planning, and its relevance is even more valuable as a driver for prevention and control of NCDs. This paper begins to address that handicap by estimating the healthcare cost of Cardiovascular Disease (Coronary Heart Disease and Stroke) in Colombia, using a person-based approach. Results show that the annual healthcare cost of a person with Coronary Heart Disease is between INT$ 4,277 and INT$ 4,846, while the cost for a person with Stroke varies between INT$5,816 and INT$6,616. The expansion of the NCDs epidemic combined with such high costs threatens the financial sustainability of health systems; primary prevention and policies targeting structural and intermediate determinants of health are a promising way to make health systems financially sustainable. [ABSTRACT FROM AUTHOR]

The modelling of deceptions in game theory and decision theory has not been well studied, despite the increasing importance of this problem in social media, public discourse, and organisational management. This paper presents an improved formulation of the extant information-theoretic models of deceptions, a framework for incorporating these models of deception into game and decision theoretic models of deception, and applies these models and this framework in an agent based evolutionary simulation that models two very common deception types employed in “fake news” attacks. The simulation results for both deception types modelled show, as observed empirically in many social systems subjected to “fake news” attacks, that even a very small population of deceivers that transiently invades a much larger population of non-deceiving agents can strongly alter the equilibrium behaviour of the population in favour of agents playing an always defect strategy. The results also show that the ability of a population of deceivers to establish itself or remain present in a population is highly sensitive to the cost of the deception, as this cost reduces the fitness of deceiving agents when competing against non-deceiving agents. Diffusion behaviours observed for agents exploiting the deception producing false beliefs are very close to empirically observed behaviours in social media, when fitted to epidemiological models. We thus demonstrate, using the improved formulation of the information-theoretic models of deception, that agent based evolutionary simulations employing the Iterated Prisoner’s Dilemma can accurately capture the behaviours of a population subject to deception attacks introducing uncertainty and false perceptions, and show that information-theoretic models of deception have practical applications beyond trivial taxonomical analysis. [ABSTRACT FROM AUTHOR]

The integration of oilfield multidisciplinary ontology is increasingly important for the growth of the Semantic Web. However, current methods encounter performance bottlenecks either in storing data and searching for information when processing large amounts of data. To overcome these challenges, we propose a domain-ontology process based on the Neo4j graph database. In this paper, we focus on data storage and information retrieval of oilfield ontology. We have designed mapping rules from ontology files to regulate the Neo4j database, which can greatly reduce the required storage space. A two-tier index architecture, including object and triad indexing, is used to keep loading times low and match with different patterns for accurate retrieval. Therefore, we propose a retrieval method based on this architecture. Based on our evaluation, the retrieval method can save 13.04% of the storage space and improve retrieval efficiency by more than 30 times compared with the methods of relational databases. [ABSTRACT FROM AUTHOR]

The number of images taken per patient scan has rapidly increased due to advances in software, hardware and digital imaging in the medical domain. There is the need for medical image annotation systems that are accurate as manual annotation is impractical, time-consuming and prone to errors. This paper presents modeling approaches performed to automatically classify and annotate radiographs using several classification schemes, which can be further applied for automatic content-based image retrieval (CBIR) and computer-aided diagnosis (CAD). Different image preprocessing and enhancement techniques were applied to augment grayscale radiographs by virtually adding two extra layers. The Image Retrieval in Medical Applications (IRMA) Code, a mono-hierarchical multi-axial code, served as a basis for this work. To extensively evaluate the image enhancement techniques, five classification schemes including the complete IRMA code were adopted. The deep convolutional neural network systems Inception-v3 and Inception-ResNet-v2, and Random Forest models with 1000 trees were trained using extracted Bag-of-Keypoints visual representations. The classification model performances were evaluated using the ImageCLEF 2009 Medical Annotation Task test set. The applied visual enhancement techniques proved to achieve better annotation accuracy in all classification schemes. [ABSTRACT FROM AUTHOR]

Transdisciplinary collaboration is the key for innovation. An evaluation mechanism is necessary to ensure that academic credit for this costly process can be allocated fairly among coauthors. This paper proposes a set of quantitative measures (e.g., t_credit and t_index) to reflect authors’ transdisciplinary contributions to publications. These measures are based on paper-topic probability distributions and author-topic probability distributions. We conduct an empirical analysis of the information retrieval domain which demonstrates that these measures effectively improve the results of harmonic_credit and h_index measures by taking into account the transdisciplinary contributions of authors. The definitions of t_credit and t_index provide a fair and effective way for research organizations to assign credit to authors of transdisciplinary publications. [ABSTRACT FROM AUTHOR]

Deep learning techniques, e.g., Convolutional Neural Networks (CNNs), have been explosively applied to the research in the fields of information retrieval and natural language processing. However, few research efforts have addressed semantic indexing with deep learning. The use of semantic indexing in the biomedical literature has been limited for several reasons. For instance, MEDLINE citations contain a large number of semantic labels from automatically annotated MeSH terms, and for a great deal of the literature, only the information of the title and the abstract is readily available. In this paper, we propose a Boltzmann Convolutional neural network framework (B-CNN) for biomedicine semantic indexing. In our hybrid learning framework, the CNN can adaptively deal with features of documents that have sequence relationships, and can capture context information accordingly; the Deep Boltzmann Machine (DBM) merges global (the entity in each document) and local information through its training with undirected connections. Additionally, we have designed a hierarchical coarse to fine style indexing structure for learning and classifying documents, and a novel feature extension approach with word sequence embedding and Wikipedia categorization. Comparative experiments were conducted for semantic indexing of biomedical abstract documents; these experiments verified the encouraged performance of our B-CNN model. [ABSTRACT FROM AUTHOR]

Background: AIDS is a worrying public health issue in China and lacks timely and effective surveillance. With the diffusion and adoption of the Internet, the ‘big data’ aggregated from Internet search engines, which contain users’ information on the concern or reality of their health status, provide a new opportunity for AIDS surveillance. This paper uses search engine data to monitor and forecast AIDS in China. Methods: A machine learning method, artificial neural networks (ANNs), is used to forecast AIDS incidences and deaths. Search trend data related to AIDS from the largest Chinese search engine, Baidu.com, are collected and selected as the input variables of ANNs, and officially reported actual AIDS incidences and deaths are used as the output variable. Three criteria, the mean absolute percentage error, the root mean squared percentage error, and the index of agreement, are used to test the forecasting performance of the ANN method. Results: Based on the monthly time series data from January 2011 to June 2017, this article finds that, under the three criteria, the ANN method can lead to satisfactory forecasting of AIDS incidences and deaths, regardless of the change in the number of search queries. Conclusions: Despite the inability to self-detect HIV/AIDS through online searching, Internet-based data should be adopted as a timely, cost-effective complement to a traditional AIDS surveillance system. [ABSTRACT FROM AUTHOR]

Few if any natural resource systems are completely understood and fully observed. Instead, there almost always is uncertainty about the way a system works and its status at any given time, which can limit effective management. A natural approach to uncertainty is to allocate time and effort to the collection of additional data, on the reasonable assumption that more information will facilitate better understanding and lead to better management. But the collection of more data, either through observation or investigation, requires time and effort that often can be put to other conservation activities. An important question is whether the use of limited resources to improve understanding is justified by the resulting potential for improved management. In this paper we address directly a change in value from new information collected through investigation. We frame the value of information in terms of learning through the management process itself, as well as learning through investigations that are external to the management process but add to our base of understanding. We provide a conceptual framework and metrics for this issue, and illustrate them with examples involving Florida scrub-jays (Aphelocoma coerulescens). [ABSTRACT FROM AUTHOR]

Systematic reviews and/or meta-analyses generally provide the best evidence for medical research. Authors are recommended to use flow diagrams to present the review process, allowing for better understanding among readers. However, no studies as of yet have assessed the quality of flow diagrams in systematic review/meta-analyses. Our study aims to evaluate the quality of systematic review/meta-analyses over a period of ten years, by assessing the quality of the flow diagrams, and the correlation to the methodological quality. Two hundred articles of “systematic review” and/or “meta-analysis” from January 2004 to August 2015 were randomly retrieved in Pubmed to be assessed for the flow diagram and methodological qualities. The flow diagrams were evaluated using a 16-grade scale corresponding to the four stages of PRISMA flow diagram. It composes four parts: Identification, Screening, Eligibility and Inclusion. Of the 200 articles screened, 154 articles were included and were assessed with AMSTAR checklist. Among them, 78 articles (50.6%) had the flow diagram. Over ten years, the proportion of papers with flow diagram available had been increasing significantly with regression coefficient beta = 5.649 (p = 0.002). However, the improvement in quality of the flow diagram increased slightly but not significantly (regression coefficient beta = 0.177, p = 0.133). Our analysis showed high variation in the proportion of articles that reported flow diagram components. The lowest proportions were 1% for reporting methods of duplicates removal in screening phase, followed by 6% for manual search in identification phase, 22% for number of studies for each specific/subgroup analysis, 27% for number of articles retrieved from each database, and 31% for number of studies included in qualitative analysis. The flow diagram quality was correlated with the methodological quality with the Pearson’s coefficient r = 0.32 (p = 0.0039). Therefore, this review suggests that the reporting quality of flow diagram is less satisfactory, hence not maximizing the potential benefit of the flow diagrams. A guideline with standardized flow diagram is recommended to improve the quality of systematic reviews, and to enable better reader comprehension of the review process. [ABSTRACT FROM AUTHOR]

This paper presents the concept of Simultaneous Localization and Multi-Mapping (SLAMM). It is a system that ensures continuous mapping and information preservation despite failures in tracking due to corrupted frames or sensor’s malfunction; making it suitable for real-world applications. It works with single or multiple robots. In a single robot scenario the algorithm generates a new map at the time of tracking failure, and later it merges maps at the event of loop closure. Similarly, maps generated from multiple robots are merged without prior knowledge of their relative poses; which makes this algorithm flexible. The system works in real time at frame-rate speed. The proposed approach was tested on the KITTI and TUM RGB-D public datasets and it showed superior results compared to the state-of-the-arts in calibrated visual monocular keyframe-based SLAM. The mean tracking time is around 22 milliseconds. The initialization is twice as fast as it is in ORB-SLAM, and the retrieved map can reach up to 90 percent more in terms of information preservation depending on tracking loss and loop closure events. For the benefit of the community, the source code along with a framework to be run with Bebop drone are made available at . [ABSTRACT FROM AUTHOR]

In this paper, we present a research work on a novel methodology of identifying abnormal behaviors at the underlying network monitor layer during runtime based on the execution patterns of Web of Things (WoT) applications. An execution pattern of a WoT application is a sequence of profiled time delays between the invocations of involved Web services, and it can be obtained from WoT platforms. We convert the execution pattern to a time sequence of network flows that are generated when the WoT applications are executed. We consider such time sequences as a whitelist. This whitelist reflects the valid application execution patterns. At the network monitor layer, our applied RETE algorithm examines whether any given runtime sequence of network flow instances does not conform to the whitelist. Through this approach, it is possible to interpret a sequence of network flows with regard to application logic. Given such contextual information, we believe that the administrators can detect and reason about any abnormal behaviors more effectively. Our empirical evaluation shows that our RETE-based algorithm outperforms the baseline algorithm in terms of memory usage. [ABSTRACT FROM AUTHOR]

Web of Things (WoT) platforms are growing fast so as the needs for composing WoT apps more easily and efficiently. We have recently commenced the campaign to develop an interface where users can issue requests for WoT apps entirely in natural language. This requires an effort to build a system that can learn to identify relevant WoT functions that fulfill user’s requests. In our preceding work, we trained a supervised learning system with thousands of publicly-available IFTTT app recipes based on conditional random fields (CRF). However, the sub-par accuracy and excessive training time motivated us to devise a better approach. In this paper, we present a novel solution that creates a separate learning engine for each trigger service. With this approach, parallel and incremental learning becomes possible. For inference, our system first identifies the most relevant trigger service for a given user request by using an information retrieval technique. Then, the learning engine associated with the trigger service predicts the most likely pair of trigger and action functions. We expect that such two-phase inference method given parallel learning engines would improve the accuracy of identifying related WoT functions. We verify our new solution through the empirical evaluation with training and test sets sampled from a pool of refined IFTTT app recipes. We also meticulously analyze the characteristics of the recipes to find future research directions. [ABSTRACT FROM AUTHOR]

Medical image collections contain a wealth of information which can assist radiologists and medical experts in diagnosis and disease detection for making well-informed decisions. However, this objective can only be realized if efficient access is provided to semantically relevant cases from the ever-growing medical image repositories. In this paper, we present an efficient method for representing medical images by incorporating visual saliency and deep features obtained from a fine-tuned convolutional neural network (CNN) pre-trained on natural images. Saliency detector is employed to automatically identify regions of interest like tumors, fractures, and calcified spots in images prior to feature extraction. Neuronal activation features termed as neural codes from different CNN layers are comprehensively studied to identify most appropriate features for representing radiographs. This study revealed that neural codes from the last fully connected layer of the fine-tuned CNN are found to be the most suitable for representing medical images. The neural codes extracted from the entire image and salient part of the image are fused to obtain the saliency-injected neural codes (SiNC) descriptor which is used for indexing and retrieval. Finally, locality sensitive hashing techniques are applied on the SiNC descriptor to acquire short binary codes for allowing efficient retrieval in large scale image collections. Comprehensive experimental evaluations on the radiology images dataset reveal that the proposed framework achieves high retrieval accuracy and efficiency for scalable image retrieval applications and compares favorably with existing approaches. [ABSTRACT FROM AUTHOR]

Public cloud storage services are becoming prevalent and myriad data sharing, archiving and collaborative services have emerged which harness the pay-as-you-go business model of public cloud. To ensure privacy and confidentiality often encrypted data is outsourced to such services, which further complicates the process of accessing relevant data by using search queries. Search over encrypted data schemes solve this problem by exploiting cryptographic primitives and secure indexing to identify outsourced data that satisfy the search criteria. Almost all of these schemes rely on exact matching between the encrypted data and search criteria. A few schemes which extend the notion of exact matching to similarity based search, lack realism as those schemes rely on trusted third parties or due to increase storage and computational complexity. In this paper we propose Oblivious Similarity based Search () for encrypted data. It enables authorized users to model their own encrypted search queries which are resilient to typographical errors. Unlike conventional methodologies, ranks the search results by using similarity measure offering a better search experience than exact matching. It utilizes encrypted bloom filter and probabilistic homomorphic encryption to enable authorized users to access relevant data without revealing results of search query evaluation process to the untrusted cloud service provider. Encrypted bloom filter based search enables to reduce search space to potentially relevant encrypted data avoiding unnecessary computation on public cloud. The efficacy of is evaluated on Google App Engine for various bloom filter lengths on different cloud configurations. [ABSTRACT FROM AUTHOR]

In recent years, there is an increasing demand for sharing and integration of medical data in biomedical research. In order to improve a health care system, it is required to support the integration of data by facilitating semantic interoperability systems and practices. Semantic interoperability is difficult to achieve in these systems as the conceptual models underlying datasets are not fully exploited. In this paper, we propose a semantic framework, called Medical Knowledge Discovery and Data Mining (MedKDD), that aims to build a topic hierarchy and serve the semantic interoperability between different ontologies. For the purpose, we fully focus on the discovery of semantic patterns about the association of relations in the heterogeneous information network representing different types of objects and relationships in multiple biological ontologies and the creation of a topic hierarchy through the analysis of the discovered patterns. These patterns are used to cluster heterogeneous information networks into a set of smaller topic graphs in a hierarchical manner and then to conduct cross domain knowledge discovery from the multiple biological ontologies. Thus, patterns made a greater contribution in the knowledge discovery across multiple ontologies. We have demonstrated the cross domain knowledge discovery in the MedKDD framework using a case study with 9 primary biological ontologies from Bio2RDF and compared it with the cross domain query processing approach, namely SLAP. We have confirmed the effectiveness of the MedKDD framework in knowledge discovery from multiple medical ontologies. [ABSTRACT FROM AUTHOR]

For the purposes of information retrieval, users must find highly relevant documents from within a system (and often a quite large one comprised of many individual documents) based on input query. Ranking the documents according to their relevance within the system to meet user needs is a challenging endeavor, and a hot research topic–there already exist several rank-learning methods based on machine learning techniques which can generate ranking functions automatically. This paper proposes a parallel B cell algorithm, RankBCA, for rank learning which utilizes a clonal selection mechanism based on biological immunity. The novel algorithm is compared with traditional rank-learning algorithms through experimentation and shown to outperform the others in respect to accuracy, learning time, and convergence rate; taken together, the experimental results show that the proposed algorithm indeed effectively and rapidly identifies optimal ranking functions. [ABSTRACT FROM AUTHOR]

Risk-based approaches have been increasingly accepted and operationalized in flood risk management during recent decades. For instance, commercial flood risk models are used by the insurance industry to assess potential losses, establish the pricing of policies and determine reinsurance needs. Despite considerable progress in the development of loss estimation tools since the 1980s, loss estimates still reflect high uncertainties and disparities that often lead to questioning their quality. This requires an assessment of the validity and robustness of loss models as it affects prioritization and investment decision in flood risk management as well as regulatory requirements and business decisions in the insurance industry. Hence, more effort is needed to quantify uncertainties and undertake validations. Due to a lack of detailed and reliable flood loss data, first order validations are difficult to accomplish, so that model comparisons in terms of benchmarking are essential. It is checked if the models are informed by existing data and knowledge and if the assumptions made in the models are aligned with the existing knowledge. When this alignment is confirmed through validation or benchmarking exercises, the user gains confidence in the models. Before these benchmarking exercises are feasible, however, a cohesive survey of existing knowledge needs to be undertaken. With that aim, this work presents a review of flood loss–or flood vulnerability–relationships collected from the public domain and some professional sources. Our survey analyses 61 sources consisting of publications or software packages, of which 47 are reviewed in detail. This exercise results in probably the most complete review of flood loss models to date containing nearly a thousand vulnerability functions. These functions are highly heterogeneous and only about half of the loss models are found to be accompanied by explicit validation at the time of their proposal. This paper exemplarily presents an approach for a quantitative comparison of disparate models via the reduction to the joint input variables of all models. Harmonization of models for benchmarking and comparison requires profound insight into the model structures, mechanisms and underlying assumptions. Possibilities and challenges are discussed that exist in model harmonization and the application of the inventory in a benchmarking framework. [ABSTRACT FROM AUTHOR]

With the recent evolution of technology, the number of image archives has increased exponentially. In Content-Based Image Retrieval (CBIR), high-level visual information is represented in the form of low-level features. The semantic gap between the low-level features and the high-level image concepts is an open research problem. In this paper, we present a novel visual words integration of Scale Invariant Feature Transform (SIFT) and Speeded-Up Robust Features (SURF). The two local features representations are selected for image retrieval because SIFT is more robust to the change in scale and rotation, while SURF is robust to changes in illumination. The visual words integration of SIFT and SURF adds the robustness of both features to image retrieval. The qualitative and quantitative comparisons conducted on Corel-1000, Corel-1500, Corel-2000, Oliva and Torralba and Ground Truth image benchmarks demonstrate the effectiveness of the proposed visual words integration. [ABSTRACT FROM AUTHOR]

Chinese herbal medicine image recognition and retrieval have great potential of practical applications. Several previous studies have focused on the recognition with hand-crafted image features, but there are two limitations in them. Firstly, most of these hand-crafted features are low-level image representation, which is easily affected by noise and background. Secondly, the medicine images are very clean without any backgrounds, which makes it difficult to use in practical applications. Therefore, designing high-level image representation for recognition and retrieval in real world medicine images is facing a great challenge. Inspired by the recent progress of deep learning in computer vision, we realize that deep learning methods may provide robust medicine image representation. In this paper, we propose to use the Convolutional Neural Network (CNN) for Chinese herbal medicine image recognition and retrieval. For the recognition problem, we use the softmax loss to optimize the recognition network; then for the retrieval problem, we fine-tune the recognition network by adding a triplet loss to search for the most similar medicine images. To evaluate our method, we construct a public database of herbal medicine images with cluttered backgrounds, which has in total 5523 images with 95 popular Chinese medicine categories. Experimental results show that our method can achieve the average recognition precision of 71% and the average retrieval precision of 53% over all the 95 medicine categories, which are quite promising given the fact that the real world images have multiple pieces of occluded herbal and cluttered backgrounds. Besides, our proposed method achieves the state-of-the-art performance by improving previous studies with a large margin. [ABSTRACT FROM AUTHOR]

The user-based collaborative filtering (CF) algorithm is one of the most popular approaches for making recommendation. Despite its success, the traditional user-based CF algorithm suffers one serious problem that it only measures the influence between two users based on their symmetric similarities calculated by their consumption histories. It means that, for a pair of users, the influences on each other are the same, which however may not be true. Intuitively, an expert may have an impact on a novice user but a novice user may not affect an expert at all. Besides, each user may possess a global importance factor that affects his/her influence to the remaining users. To this end, in this paper, we propose an asymmetric user influence model to measure the directed influence between two users and adopt the PageRank algorithm to calculate the global importance value of each user. And then the directed influence values and the global importance values are integrated to deduce the final influence values between two users. Finally, we use the final influence values to improve the performance of the traditional user-based CF algorithm. Extensive experiments have been conducted, the results of which have confirmed that both the asymmetric user influence model and global importance value play key roles in improving recommendation accuracy, and hence the proposed method significantly outperforms the existing recommendation algorithms, in particular the user-based CF algorithm on the datasets of high rating density. [ABSTRACT FROM AUTHOR]

A commonly used strategy to improve search accuracy is through feedback techniques. Most existing work on feedback relies on positive information, and has been extensively studied in information retrieval. However, when a query topic is difficult and the results from the first-pass retrieval are very poor, it is impossible to extract enough useful terms from a few positive documents. Therefore, the positive feedback strategy is incapable to improve retrieval in this situation. Contrarily, there is a relatively large number of negative documents in the top of the result list, and it has been confirmed that negative feedback strategy is an important and useful way for adapting this scenario by several recent studies. In this paper, we consider a scenario when the search results are so poor that there are at most three relevant documents in the top twenty documents. Then, we conduct a novel study of multiple strategies for relevance feedback using both positive and negative examples from the first-pass retrieval to improve retrieval accuracy for such difficult queries. Experimental results on these TREC collections show that the proposed language model based multiple model feedback method which is generally more effective than both the baseline method and the methods using only positive or negative model. [ABSTRACT FROM AUTHOR]