Search

In this research, the well-known microblogging site, Twitter, was used for a sentiment analysis investigation. We propose an ensemble learning approach based on the meta-level features of seven existing lexicon resources for automated polarity sentiment classification. The ensemble employs four base learners (a Two-Class Support Vector Machine, a Two-Class Bayes Point Machine, a Two-Class Logistic Regression and a Two-Class Decision Forest) for the classification task. Three different labelled Twitter datasets were used to evaluate the effectiveness of this approach to sentiment analysis. Our experiment shows that, based on a combination of existing lexicon resources, the ensemble learners minimize the error rate by avoiding poor selection from stand-alone classifiers.

Purpose The purpose of this paper is to introduce a model for identifying, storing and sharing contextual information across smartphone apps that uses the native device services. The authors present the idea of using user input and interaction within an app as contextual information, and how each app can identify and store contextual information. Design/methodology/approach Contexts are modeled as hierarchical objects that can be stored and shared by applications using native mechanisms. A proof-of-concept implementation of the model for the Android platform demonstrates contexts modelled as hierarchical objects stored and shared by applications using native mechanisms. Findings The model was found to be practically viable by implemented sample apps that share context and through a performance analysis of the system. Practical implications The contextual data-sharing model enables the creation of smart apps and services without being tied to any vendor’s cloud services. Originality/value This paper introduces a new approach for sharing context in smartphone applications that does not require cloud services.

The goal of this project was to demonstrate the possibilities of open source search engine and aggregation technology in a Web environment by building a meta-search engine which employs free open search engines and open protocols. In contrast many meta-search engines on the Internet use proprietary search systems. The search engines employed in this case study are all based on the OpenSearch protocol. OpenSearch-compliant systems support XML technologies such as RSS and Atom for aggregation and distribution. The meta-search engine itself combines the ranked lists of the chosen search sources based on user-supplied weightings. This is implemented in Lucene, a free open source information retrieval library.

A software frameworks is a way of tackling the issue of software re-usability. The concept of a software framework is introduced and we design a general-purpose framework for probabilistic information retrieval. We discuss software engineering aspects of retrieval systems in general and previous efforts at building re-usable IR systems. Design patterns are utilised throughout the framework. Issues associated with a C++ implementation are also touched upon.

Web syndication has become a popular means of delivering relevant information to people online but the complexity of standards, algorithms and applications pose considerable challenges to engineers. This paper describes the design and development of a novel Web-based syndication intermediary called InterSynd and a simple Web client as a proof of concept. We developed format-neutral middleware that sits between content sources and the user. Additional objectives were to add feed discovery and recommendation components to the intermediary. A search-based feed discovery module helps users find relevant feed sources. Implicit collaborative recommendations of new feeds are also made to the user. The syndication software built uses open standard XML technologies and the free open source libraries. Extensibility and re-configurability were explicit goals. The experience shows that a modular architecture can combine open source modules to build state-of-the-art syndication middleware and applications. The data produced by software metrics indicate the high degree of modularity retained.

The goal of this project was to demonstrate the possibilities of open source search engine and aggregation technology in a Web environment by building a meta-search engine which employs free open search engines and open protocols. In contrast many meta-search engines on the Internet use proprietary search systems. The search engines employed in this case study are all based on the OpenSearch protocol. OpenSearch-compliant systems support XML technologies such as RSS and Atom for aggregation and distribution. The meta-search engine itself combines the ranked lists of the chosen search sources based on user-supplied weightings. This is implemented in Lucene, a free open source information retrieval library.

We present here an overview of a research project which is ongoing at University College Cork. Information filtering software is being constructed which interfaces to the USENET Net news utility and which screens out irrelevant articles prior to their being presented to a user. The system is based on an intelligent agent approach and embodies machine learning, adaptation and relevance feedback techniques in its construction. A weighed graph representation is used for articles, and graph manipulation algorithms are used in the processing.

The growth of online health communities, particularly those involving socially generated content, can provide considerable value for society. Participants can gain knowledge of medical information or interact with peers on medical forum platforms. However, the sheer volume of information so generated – and the consequent ‘noise’ associated with large data volumes – can create difficulties for information consumers. We propose a solution to this problem by applying high-level analytics to the data – primarily sentiment analysis, but also content and topic analysis - for accurate classification. We believe that such analysis can be of significant value to data users, such as identifying a particular aspect of an information space, determining themes that predominate among a large dataset, and allowing people to summarize topics within a big dataset. In this thesis, we apply machine learning strategies to identify sentiments expressed in online medical forums that discuss Lyme Disease. As part of this process, we distinguish a complete and relevant set of categories that can be used to characterize Lyme Disease discourse. We present a feature-based model that employs supervised learning algorithms and assess the feasibility and accuracy of this sentiment classification model. We further evaluate our model by assessing its ability to adapt to an online medical forum discussing a disease with similar characteristics, Lupus. The experimental results demonstrate the effectiveness of our approach. In many sentiment analysis applications, the labelled training datasets are expensive to obtain, whereas unlabelled datasets are readily available. Therefore, we present an adaptation of a well-known semi-supervised learning technique, in which co-training is implemented by combining labelled and unlabelled data. Our results would suggest the ability to learn even with limited labelled data. In addition, we investigate complementary analytic techniques – content and topic analysis – to leverage best used of the data for various consumer groups. Within the work described in this thesis, some particular research issues are addressed, specifically when applied to socially generated medical/health datasets: • When applying binary sentiment analysis to short-form text data (e.g. Twitter), could meta-level features improve performance of classification? • When applying more complex multi-class sentiment analysis to classification of long-form content-rich text data, would meta-level features be a useful addition to more conventional features? • Can this multi-class analysis approach be generalised to other medical/health domains? • How would alternative classification strategies benefit different groups of information consumers?

This paper describes a software development tool for automating the process of source code querying of Java programs together with an initial evaluation. It contains a catalogue of source code queries and streamlines the querying process. The GUI-driven Eclipse plugin called PwSOUL enables Java programmers to more easily check their source code for potential bugs, and find common programming idioms and design patterns. The tool should be useful to software developers as an aid for tasks such as debugging, refactoring, and software evolution and maintenance. A key aim of this work was to cut the effort required by a developer to run a suite of source code queries by simplifying and automating user actions. PwSOUL simplifies and automates aspects of Barista, an existing Eclipse plugin for Java runtime source code querying and query scheduling that uses an example-based extension of a program query language called SOUL. The main features of PwSOUL are a catalogue of 32 source code queries, and a GUI to aid navigation and automate the process of running queries and returning results.