Search

In the intelligent transformation of the mechanical manufacturing, a large amount of spare parts models is inevitably generated, which brings huge challenges to the efficient retrieval of data. Considering the user-friendly and lightweight characteristics of design sketches, this paper conducted model retrieval of mechanical parts based on design sketches by constructing a deep cross-domain representation model. Aiming at the problem of cross-modal information association between sketches and 3D models, this paper proposed a feature joint learning method, which controlled the differences within and between classes of retrieved models. By this method, feature descriptors could acquire single-domain features as well as fuse the cross-domain information, and thus, the method established a consistent association representation of cross-modal data in a co-embedding space. Finally, it constructed the index table by using hash coding to realize fast retrieval of massive data. The experimental results on the parts data show that the proposed retrieval method based on the design sketch can not only achieve the most accurate retrieval results, but also have higher retrieval efficiency among the comparison methods. The method not only improves the accuracy of cross-modal spare parts information retrieval, but also improves the efficiency of data management, thereby indirectly improving the efficiency and convenience of product design. The related systems have been applied in some enterprises and have received good feedback. [ABSTRACT FROM AUTHOR]

Aiming at the traditional ciphertext retrieval scheme in cloud environment, which generates file vectors and retrieval vectors based on statistical model, and does not consider the deep-seated semantic information of files and requests, this paper proposes a deep-seated semantic ciphertext retrieval model based on hybrid Cloud Architecture. The vector generation model is constructed through the private cloud federated learning neural network model, and the ciphertext data is stored through the public cloud. In addition, this paper proposes a secret inverted index table to store file vectors, so as to improve the efficiency of retrieval without ensuring that the retrieval information is not leaked in the retrieval process of public cloud. The analysis and experiments on real data sets show that this scheme is better than the current ciphertext retrieval schemes of the same type in terms of security and search efficiency. [ABSTRACT FROM AUTHOR]

Objectives: In order to improve the calculation efficiency of linear octree neighbor grid elements, this paper uses Hilbert code to mark the grid elements and proposes a fast calculation algorithm for neighbor grid elements Hilbert code. Methods: Based on the Hilbert primitive curve, the concept of state vector is introduced to record the filling order of the Hilbert curve to all child elements belonging to the same parent element, thereby establishing the hierarchical evolution and degradation function of the state vector. With the hierarchical mapping relationship in the order curve, the state vector and its hierarchical evolution function are finally used to calculate the neighboring grid element Hilbert code. Results: The results of comparison experiments on calculation accuracy and efficiency show that the calculation result of the algorithm is correct. The calculation speed of the state vector decreases with the increase of the level. The calculation of 4 201 grid elements can be completed within 1 ms on the 20th level. In the calculation of the same number of neighbor grid elements at the specified level, the speed of this algorithm is significantly better than the existing Morton code conversion algorithm. At the 15th level, the speed of the algorithm in this paper is about 2.1-2.4 times that of the existing Morton code conversion algorithm. In the calculation of millions of adjacent cells at different levels, the calculation speed of the algorithm in this paper is compared with the existing Morton code conversion. The improvement factor of the algorithm increases as the level increases. At level 20, the efficiency of the proposed algorithm increases by 2.6 times. Conclusions: Experimental results show that this algorithm is superior to the existing Morton code conversion algorithm, and can be used as the basis of spatial operations such as efficient data retrieval, determination of three-dimensional topological relationship and connectivity judgment. [ABSTRACT FROM AUTHOR]

Privacy information security retrieval of multi-party data sources is an urgent problem in network security. The feature of oblivious transfer technology is to ensure the data security of each participant. Therefore, this paper combined oblivious transfer technology with homomorphic encryption and symmetric encryption in cryptography to design an anonymous query protocol for multi-party data sources. Firstly, based on the idea of oblivious transfer, this paper designed a three-party anonymous query protocol, and gave the model of the protocol and the specific process of the protocol. Secondly, this paper proved and analyzed the correctness and security of the protocol. Finally, it extended the three-party anonymous query protocol to the multiparty query protocol, and it applied to the verification of blacklist users in digital product transaction, which solved the problem of malicious transaction of buyer users in the process of network security transaction. Experimental data shows that the algorithm has correct query results and high efficiency under the condition of ensuring the data security of all parties. [ABSTRACT FROM AUTHOR]

The success of open source software promoted the vigorous development of the software industry, and a large number of code resources have created conditions for code search. How to find the required code through code search technology has become an important issue. In order to better promote the follow-up research, firstly this paper introduced and explained the concept and research trend of code search. Secondly, it reviewed the research of code search using different technologies, including information retrieval, query and code description enhancement, program feature classification, and deep learning, and summarized the differences. Next, it introduced the application of code search technology in many fields, including program synthesis, code recommendation and completion, and code style improvement. Finally, this paper analyzed the main problems of code search in order to provide some references and suggestions for the future development of this direction. [ABSTRACT FROM AUTHOR]

An important measure to ensure the national food security and improve the quality of arable land is to demarcate the prime farmland scientifically and reasonably. There are three types of evaluation units for delimiting prime farmland, mainly including administrative villages, plots and grids. Compared with the other two types, grid can not only break the constraints of irregular geographic data, improve the ability of data update and retrieval, but also integrate, share and use information resources, which is easy to locate in the space. In the current research field, the demarcation of prime farmland is generally carried out in a uniform grid environment. If using large-scale uniform grid, however, the area of small and scattered arable land will have a large error. Although using small-scale uniform grid can reduce the area error, there are other problems such as large amount of data and high redundancy. The advantages of non-uniform grid are obvious. Its data volume is far less than that of uniform grid, which effectively reduces the frequency of invalid data search. And, the efficiency of data retrieval and transmission can be improved. “Centralized and continuous, high quality” is the requirement of realizing the construction of prime farmland with high spatial standard. Therefore, the purpose of this paper is to establish a method to divide and express prime farmland by using arable land concentricity and arable land quality index with the non-uniform grid environment. Wenxi County of Shanxi Province is taken as the research area. The research methods of the paper are as follows: to construct non-uniform grid and multi-scale uniform grid with the help of Arc-Engine secondary development platform; Comparing the expression error of arable land area and the data amount of grid, and then analyzing the advantage of non-uniform grid in cultivated land expression. Focusing on the contiguous land and the quality of cultivated land evaluates the spatial distribution of cultivated land. The results show that the arable land is divided into permanent protection type, quality improvement type, key management type and reserve management type by the method of spatial superposition; On the basis of the connotation of prime farmland, demarcating and expressing prime farmland. Using non-uniform grid can not only reduce data redundancy, but also achieve small error expression of arable land information. The delimitation results of prime farmland in Wenxi County meet the requirements of “centralized, continuous and high quality”. The area of prime farmland accounts for 83.22% of the total area of farmland, and the demarcation proportion conforms to the national regulations. The non-uniform grid can well express the spatial distribution of prime farmland. Obviously, this method is suitable for the demarcation of prime farmland and provides a new way for expressing prime farmland better. [ABSTRACT FROM AUTHOR]

K-ary search tree is a promising structure due to its friendliness to cache awareness and SIMD acceleration. Recent years have witnessed its successful application to inverted index for efficient query processing and index compression. However, its query processing methods are either complicated or rudimentary, and its adoption is limited in membership test, which both impair its application in real life. This paper studied the intersection algorithm for k-ary search tree, and introduced two early termination techniques collaborating with particular traversal methods to improve its intersection efficiency. The experimental analyses validated the proposed optimizations could greatly improve the query speed, and the optimized k-ary search tree suits better for long inverted lists, and when equipped with larger SIMD registers, it is able to beat the state-of-the-art sorted-array based intersection algorithms. [ABSTRACT FROM AUTHOR]

In order to solve the problem of inefficient query in relational database keyword search scheme based on schema graph, this paper proposed combined candidate network evaluation( CCNE) method. This method designed a new combined candidate network(CCN) structure, which could effectively avoid the redundant operation caused by the repeated structure between candidate networks(CN) in the traditional method. At the same time, it proposed an improved CN generation strategy to improve the efficiency by avoiding generating redundant CN and narrowing the traversal range. Finally, based on CCN, this paper designed a CCN evaluation algorithm, which greatly reduced the complex database query operations required by traditional methods and further improved the query efficiency. Experiment based on real data sets show that CCNE method can improve query efficiency without any loss of query results. [ABSTRACT FROM AUTHOR]

In order to improve the accuracy of open source software bug localization, this paper proposed an open source soft. ware bug localization method based on twice propagation algorithm (TPA). It combined the restart random walk with the label propagation algorithm based on the information retrieval method. The specific location of the bug is analyzed from the three aspects: the call relationship, module risk and historical bug report of each module in the open source software. The bug localization is divided into two parts: risk propagation and label propagation. Through the similarity and calling relationship between the software modules, combined with the history of modified reports, it evaluated the module' s risk. It constructed the bug label through the modified bug reports, and propagated the software label from it. In the Eclipse 3.1 bug report data set, the top-N rank and MRR indicators are higher than the similar bug location methods. Experiments show that the TPA method has higher accuracy. [ABSTRACT FROM AUTHOR]

To solve the deficiency of Wu-Manber multi-pattern matching algorithm in terms of the achieved matching efficiency and jump distance, this paper proposed an improved Wu-Manber algorithm for Chinese PDF document content review on the basis of the coding formats of PDF document content. By employing the Bloom filter to extract the crucial information of the pattern string, and exploiting the double hash and PDF document encoding rules, the proposed improved algorithm was able to reduce the number of unnecessary matches and increase the jump distance, which results in improving the matching efficiency for the content retrieval of PDF document. The practical experimental results confirm the improved matching efficiency for PDF document, especially when the shortest mode string is long and the mode string size is large, the matching efficiency can be even doubled [ABSTRACT FROM AUTHOR]

Subsequence query technique has important applications in several fields such as finance, commerce and healthcare. However, due to the high time complexity of similarity comparison algorithms such as dynamic time warping ( DTW), the length of subsequence has a great influence on the retrieval time, which limits the efficiency of long subsequence retrieval on data sets. This paper proposed a fast subsequence query algorithm based on short sequence grouping and assembling strategy . This algorithm first separated all subsequences which were in a given length in the data set into groups and marked out the representative subsequence for each group. Then it cut the query sequence into small query sequences with a fixed length during the query processing, and used DTW algorithm to compute the candidate sets of subsequences . The representative subsequence had a high similarity with the small query sequences. Finally,it assembled all the sequences in the candidate sets to derive the query result sequences. Experiments show that the efficiency of the new algorithm is about 10 times higher than that of the typical algorithm. [ABSTRACT FROM AUTHOR]

Efficient retrieval is one of the crucial services of digital library, and sorting is the core issue of efficient information retrieval. Given a list of candidate book titles, the sorting model is used to generate a sorted list of book titles. When learning-based sorting algorithms are applied in the filed of information retrieval, they often optimize the sorting model by minimizing the value of the pairwise loss function. However, existing analysis shows that minimizing the pairwise loss value does not necessarily lead to the optimal sorting performance of the listwise algorithm. It is also very difficult to combine the online learning sorting algorithm with the listwise algorithm. This paper proposes an online learning sorting algorithm based on listwise, which aims to realize the online learning sorting algorithm under the premise of ensuring the performance advantage of listwise algorithm, thereby reducing the retrieval complexity. Firstly, the problem of combining the online learning sorting algorithm with the listwise algorithm is solved. Secondly, the sorting model is optimized by minimizing the loss function based on the predicted list and the real list. Finally, an adaptive learning rate based on the online-listwise algorithm is proposed. Experimental results show that the proposed algorithm has better retrieval performance and speed. [ABSTRACT FROM AUTHOR]

In traditional and relational database, it is difficult to meet the needs of data storage and quick retrieval due to such huge amount of data. For this problem, this paper studied the mapping model of data file field, the relationship between data file field and HBase column, and the importing execution plan, which had solved the heterogeneity of file object and the universality of storage transformation. It put forward the "RowKey" generation rules and algorithm, and provided the algorithm of data transformation based on mapping model and "RowKey". Then according to the mapping rules of data file fields with HBase table column, the data in data file was transformed into HBase. It realized the fast data access and retrieval according to prefix matching in "RowKey" or keyword matching, which had strong commonality and could be widely used in HBase large data storage applications. [ABSTRACT FROM AUTHOR]

Food safety has become an important topic in people’s daily lives. In recent years, the frequent occurrence of food safety incidents have caused a crisis of trust between consumers and the food industry. Accordingly, there is an urgent need to establish an efficient and trusted traceability system for agricultural products. In the practical application of the traditional traceability system, the data of the production, processing, packaging, transportation, storage and sales of agricultural products are stored in the local database of each node, data in traditional traceability systems is centralized, and authoritative agencies manage the central database of the traceability system. Because the traceability data of each supply chain node is managed by the enterprise itself, the data is easy to be tampered with. Therefore, it is necessary to improve the reliability of information transmission between different roles in agricultural supply chain.Blockchain has the characteristics of decentralization, non-tampering, traceability, and high availability. Using these characteristics of blockchain and using blockchain technology to trace agricultural products, it is possible to solve the problems existing in the current traditional traceability system.At present, Scholars in China and abroad have studied the application of blockchain in the field of traceability. However, the traceability system of blockchain have some problems, such as low query efficiency, high data-storage load, and poor data privacy and security. In order to solve the above problems, this study designed a storage structure for a blockchain-based traceability system for agricultural products based on Hyperledger Fabric and proposed an index storage and query method for gtraceability information. In this method, “On chain and off chain” data storage technology is used, the enterprise first digitally signs the hash value of the traceability information of agricultural products, writes the hash value into the blockchain, and returns the block number to the blockchain where the hash value is located. Then, the traceability information of agricultural products and the block number are stored in the local database of each node. If the information of agricultural products needs to be modified, it is necessary to rewrite the hash value of the traceability information to the blockchain and update the block number. When consumers scan the QR code of agricultural products, the traceability information and block number is obtained from the local database, the value is calculated by hash algorithm for the acquired traceability information, and compare it with this newly acquired hash value with the hash value stored in the blockchain by block number to verify the authenticity of traceability information of agricultural products. The result showed that, when the number of total traceability records reached 11×104 in the blockchain and the number of batch traceability records was 400 respectively, the query efficiency of the newly proposed method was increased by 70.56% and 88.66% compared to two traditional blockchain-based data query methods of query performed using the value of key attribute and querying performed using the batch number of agricultural products through the rich query of the CouchDB database, respectively. The approach of this paper effectively solves the problem of excessive load pressure of blockchain data storage, ensures data privacy and security, and improves the reliability and timeliness of traceability information. [ABSTRACT FROM AUTHOR]

The bug localization of large software system usually adopted the method of text similarity based on source codes and bug reports, but this method couldn't make full use of the source code structure information. In order to solve this prob-lem, this paper proposed a method for bug localization based on source code structured information (such as class names, method names, variable names, comments, etc.)and historical bug report information. It gave different weight for different parts of source code structure when it matched the bug report. In addition, the method also took the advantage of historical bug report information to improve the accuracy of bug localization. In order to verify the effectiveness of the method, this paper conducted a case study using the Eclipse project, and compared the results of the method with BugLocator and BugScout. The experimental results show that the method is better than BugLocator and BugScout in efficiency and accuracy of the bug locali-zation. [ABSTRACT FROM AUTHOR]

After analyzing platform-Spark and the projection tree algorithm, this paper proposed the projection tree algorithm based on Spark( PTBS). In view of the shortcomings of the serial algorithm in dealing with large-scale data, the method had been improved in many aspects: to reduce the cost caused by the support count of candidate itemsets, changed the data storage structure; data was grouped by prefix and divided into different computing nodes to compress the searching space of data and realized parallel computation. Finally it implemented the algorithm on the Spark platform. Experiments show that the algorithm can run efficiently when dealing with massive data, and has good expansibility in the face of large scale data growth. [ABSTRACT FROM AUTHOR]

In the traditional Web page ranking algorithm Okapi BM25, there exists a problem that the retrieval results are independent to the domain keywords, and the improved algorithm needs to build the domain vector manually. To address this issue, this paper proposed a Web page ranking algorithm based on BM25 and softmax regression classification model. The method first encoded the Web page text with the bag-of-words model. And then trained the softmax regression classification model by a small amount of Web data to predict the category scores of the test Web data. Finally it combined the category scores and the BM25 information retrieval scores to get the final ranking of Web page results. Experimental results show that this method can meet the user’s information need better without even manually creating the domain vector. [ABSTRACT FROM AUTHOR]

Distributed information retrieval is an important research issue in information retrieval field. In order to improve the performance of distributed information retrieval, this paper proposed a document replication locality based distributed information retrieval method. For any site, if the remote documents, which are in the result set, are replicated in local site, then the query forwards between sites can be reduced, and thus improve the performance of information retrieval. Based on this idea, this paper transformed the replication placement of distributed information retrieval into the locality of query, built corresponding optimization model, and proposed corresponding replication placement strategies for different replication placement models. The simulation experiments show that, the proposed method is more accurate, and has less response time compared with related work. [ABSTRACT FROM AUTHOR]

From the accessibility perspective of the journal Web citation, the paper researches the variation of information resources, it may benefit the development of traditional citation databases. This paper analyzes the journal Web citations inaccessibility research status at home and abroad, investigates the causes of this phenomenon, and presents major solutions. It points out that journal web citation accessibility system platform should focus on solving the variation of information resources. This platform may include the retrieval modules of current and archive issues, reconstruct journal web citation access mechanism so as to realize the maximum retrieval and presenting of Web citations. [ABSTRACT FROM AUTHOR]

With the development of network technology, more and more resources are applied in information retrieval in the Internet. Numerous studies show that the social annotation can be used to improve search quality. In the existing personalized ranking methods, the similarity between users is usually calculated by their commonly used tag sets. However, in reality, there are some problems such as the sparseness of user annotation data and label synonyms, which makes the similarity calculation inaccurate. Based on the previous researches, this paper proposes a personalized ranking method for fusing the similar topic domains. Firstly, this method separates the webpage and tags with different thematic meanings, and finds the tag synonyms by constructing the network of similar tags. Secondly, this method finds the users of similar interests by combing the user's tags and the preference of topic domains, and extends the user's tag information to improve the personalized information retrieval effectively. Experimental results on real data show that this method can effectively alleviate the problems of data sparsity and tag synonyms, and can help to improve the user's search experience. [ABSTRACT FROM AUTHOR]

Aiming al the retrieval semantic information in Mongolian, this paper proposed a new method combined topic model latent dirichlet allocation (LDA) and language model. This method modeled Mongolian documents with LDA and language model, estimated parameters with Gibbs sampling and represented probability of word, it could mine the hidden relationship between the different topics and the words from documents, got the topic distribution and computed the similarity of keywords topics. Finally, it returned to the most relevant documents with topics. Experimental results show that the method has a higher performance in topic semantic compared with one sole model. [ABSTRACT FROM AUTHOR]

Online forums contains much useful information,which makes it convenient for users to retrieve necessary knowledge, however, the hierarchical structure of forum data poses great challenges to content retrieve. In order to solve this problem, this paper proposed a hierarchical scoring function based multi-granularity searching method. Firstly, it represented the forum data with trees, and gave a scoring function including topics, posts, sentences and words based on several considerations. Secondly, in order to avoid the replication of data in results of multi-granularity, it proposed a maximization model of results with constraints. Finally, it transformed the maximization model of results into the problem of maximal independent sets, and gave a heuristic optimal algorithm. The experiments show that, the proposed method is more efficient and accurate that related works while retrieving forum data. [ABSTRACT FROM AUTHOR]

Appropriate Web content extraction technique can remove the data which is redundant, repetitive and useless from massive Web pages while extracting more meaningful and more useful data. Through the observation of Web pages, this paper proposed and implemented a Web content extraction method based on the layout similarity that the pages under the same Web site showed similar in content layout and style structure. It achieves the purpose of main content extraction by comparing the similarity of the DOM node structure data from the Web pages belong to the same topic of the same sites. It also did some tentative research and implementation on some other content relevant to this content extraction method. Experiments prove that this method is simple, practical and universal, and it can not only meet the requirement of both high accuracy but also provide support for more Internet applications of content analysis. [ABSTRACT FROM AUTHOR]

Traditional topic model assumes that each document is related only one topic, but in practice, a document is usually related with multi topics. This paper represented a document with combination of multi topics with the LDA model, and proposed a LDA based mixture model for Ad hoc information retrieval method in language model framework. The proposed method combined the LDA and the document model. Compared with cluster-based model, it has a higher performance, while keeping low computation complexity, and thus can be used in information retrieval of large scale document collections. [ABSTRACT FROM AUTHOR]

Against the shortage of the query performance of information retrieval system, this paper proposed a query expansion method based on the integration of correlation rules and WordNet ontology. This method used correlation rule mining and WordNet ontology to build a weighted word graph. It calculated the importance of each word according to the graph structure and the weights. In the query stage, it selected the neighbor words of original query as the expansion word source, and then used the top p words with maximum weights for second retrieval. On the basis of the algorithm implemention, it used a full-text search machine Lucene to do experiment. At last, it compared the proposed method results F1 values with the other algorithms. Experimental results demonstrate that the proposed approach achieves 16.93% retrieval improvement compared with the method without query expansion. [ABSTRACT FROM AUTHOR]

Adaptation of ranking model according to the query history could give a better personalized query results. In order to improve the precision of information retrieval, this paper proposed a linear regression based ranking algorithm. Firstly, it proposed an adaptive general ranking framework based on linear regression. Secondly, by introducing the RankSVM algorithm into the proposed framework, proposed an adaptive RankSVM algorithm. Finally, it validated the efficiency of the proposed algorithm using real dataset experiments. [ABSTRACT FROM AUTHOR]