Jin-Cheon Na

Aspect-based sentiment analysis of movie reviews on discussion boards

In this article, a method for automatic sentiment analysis of movie reviews is proposed, implemented and evaluated. In contrast to most studies that focus on determining only sentiment orientation (positive versus negative), the proposed method performs fine-grained analysis to determine both the sentiment orientation and sentiment strength of the reviewer towards various aspects of a movie. Sentences in review documents contain independent clauses that express different sentiments toward different aspects of a movie. The method adopts a linguistic approach of computing the sentiment of a clause from the prior sentiment scores assigned to individual words, taking into consideration the grammatical dependency structure of the clause. The prior sentiment scores of about 32,000 individual words are derived from SentiWordNet with the help of a subjectivity lexicon. Negation is delicately handled. The output sentiment scores can be used to identify the most positive and negative clauses or sentences with respect to particular movie aspects.

Collaborative Querying through a Hybrid Query Clustering Approach

Harnessing previously issued queries to facilitate collaborative querying is an approach that can help users in digital libraries and other information systems better meet their information needs. Here, the kernel step is to identify and cluster similar queries by mining the query logs. However because of the short lengths of queries, it is relatively difficult to cluster queries effectively using on the terms used since they cannot convey enough information. This paper introduces a hybrid method to cluster queries by utilizing both the query terms and the results returned to queries. Experiments show that this method outperforms existing query clustering techniques.

Sentiment lexicons for health-related opinion mining

Opinion mining consists in extracting from a text opinions expressed by its author and their polarity. Lexical resources, such as polarized lexicons, are needed for this task. Opinion mining in the medical domain has not been well explored, partly because little credence is given to patients and their opinions (although more and more of them are using social media). We are interested in opinion mining of user-generated content on drugs/medication. We present in this paper the creation of our lexical resources and their adaptation to the medical domain. We first describe the creation of a general lexicon, containing opinion words from the general domain and their polarity. Then we present the creation of a medical opinion lexicon, based on a corpus of drug reviews. We show that some words have a different polarity in the general domain and in the medical one. Some words considered generally as neutral are opinionated in medical texts. We finally evaluate the lexicons and show with a simple algorithm that using our general lexicon gives better results than other well-known ones on our corpus and that adding the domain lexicon improves them as well.

Impact of online resources on informal learners: Parents' perception of their parenting skills

The Internet and Web technology development have opened up new ways for people to communicate, gain new information and increase their knowledge. One particular area of interest is that of using online resources to empower informal learners to increase their knowledge at their own time and space. One could look at an online resource as a public library being brought to the homes of such informal learners. The goal of the study described in this paper was to present evidence of the impact of online resources on such informal learners. In this study, the informal learners were a group of parents with young children aged between 0 and 6, and the knowledge being learned informally was that of their knowledge of and attitudes to their parenting skills. This study used an online parenting portal, KidzGrow Online, to identify if and how an online resource could impact on the group of parents with regard to the time they spent with their children, their own perceived level of knowledge of their childrens development, and the level of their self-confidence in their parenting role. The study showed that after a period of three months, significant differences could be observed between the responses from parents who had access to the online resource and those from parents who did not.

Dynamic documents: authoring, browsing, and analysis using a high-level petri net-based hypermedia system

caT (for Context-Aware Trellis) was initially developed to support context-aware documents by incorporating high-level Petri-net specification, context-awareness, user modeling, and fuzzy knowledge handling features into Trellis, a Petri-net-based hypermedia system. The browsing behavior of documents specified in the caT model can reflect the readers contextual (such as location and time) and preference information. Recently, to provide a framework for the authoring, browsing, and analysis of reasonably complex, dynamic documents, we added (or extended) several features in the caT system, providing hierarchical Petri net support, a structured authoring tool, browsing tools for multiple presentations of a particular documents specification, and a Petri net analysis tool. In this paper, we present the extended features of caT and give examples of using caT to define and present various documents, such as formal specification of software requirements and customized Web documents. Since caT is based on a formal model, the behavioral characteristics of developed caT models can be analyzed. Current debugging and analysis tools, integrated into the authoring tool, are also introduced.

Sentence-level sentiment polarity classification using a linguistic approach

Recent sentiment analysis research has focused on the functional relations of words using typed dependency parsing as this provides a refined analysis on the grammar and semantics of the textual data, which could improve performance. However, typed dependencies only provide the grammatical relationships between individual words while there exist more complex relationships between words that could influence a sentence sentiment polarity. In this paper, we propose a linguistic approach, called Polarity Prediction Model (PPM), that combines typed dependencies and subjective phrase analysis to detect sentence-level sentiment polarity. Our approach also considers the intensity of words and domain terms that could influence the sentiment polarity output. PPM is shown to provide a fine-grained analysis for handling and explaining the complex relationships between words in detecting a sentence sentiment polarity. PPM was found to consistently outperform a baseline model by 5% in terms of overall F1-score, and exceeding 10% in terms of positive F1- score when compared to a Typed-dependency only approach.

Phrase-level sentiment polarity classification using rule-based typed dependencies and additional complex phrases consideration

The advent of Web 2.0 has led to an increase in user-generated content on the Web. This has provided an extensive collection of free-style texts with opinion expressions that could influence the decisions and actions of their readers. Providers of such content exert a certain level of influence on the receivers and this is evident from blog sites having effect on their readers’ purchase decisions, political view points, financial planning, and others. By detecting the opinion expressed, we can identify the sentiments on the topics discussed and the influence exerted on the readers. In this paper, we introduce an automatic approach in deriving polarity pattern rules to detect sentiment polarity at the phrase level, and in addition consider the effects of the more complex relationships found between words in sentiment polarity classification. Recent sentiment analysis research has focused on the functional relations of words using typed dependency parsing, providing a refined analysis on the grammar and semantics of textual data. Heuristics are typically used to determine the typed dependency polarity patterns, which may not comprehensively identify all possible rules. We study the use of class sequential rules (CSRs) to automatically learn the typed dependency patterns, and benchmark the performance of CSR against a heuristic method. Preliminary results show CSR leads to further improvements in classification performance achieving over 80% F1 scores in the test cases. In addition, we observe more complex relationships between words that could influence phrase sentiment polarity, and further discuss on possible approaches to handle the effects of these complex relationships.

Analysis of the macro-level discourse structure of literature reviews

Purpose – The purpose of this study is to analyze the macro‐level discourse structure of literature reviews found in information science journal papers, and to identify different styles of literature review writing. Although there have been several studies of human abstracting, there are hardly any studies of how authors construct literature reviews.Design/methodology/approach – This study is carried out in the context of a project to develop a summarization system to generate literature reviews automatically. A coding scheme was developed to annotate the high‐level organization of literature reviews, focusing on the types of information. Two sets of annotations were used to check inter‐coder reliability.Findings – It was found that literature reviews are written in two distinctive styles, with different discourse structures. Descriptive literature reviews summarize individual papers/studies and provide more information on each study, such as research methods, results and interpretation. Integrative liter...

Influence detection between blog posts through blog features, content analysis, and community identity

Purpose – This study aims to investigate three common approaches – quantitative blog features analysis, content analysis, and community identification – to detect influence in the blogosphere (i.e. among blog posts).Design/methodology/approach – Quantitative analysis of blog features, together with manual sentiment and agreement analysis and community identification, were performed on blog postings and their content. Correlation studies of the selected influential variables were conducted to determine the effectiveness of each variable.Findings – Agreement expressed by the linking blogger with the linked blogger, similar sentiments expressed by both bloggers on common topics, and community identity are statistically significant features for detecting influence in the linked blogs.Research limitations/implications – A small data set of 196 blog posting pairs was used for the study as the blog features and content are analysed manually. Nonetheless statistical analysis on the data set identified significant...

Comparing sentiment expression in movie reviews from four online genres

Purpose – This paper aims to investigate the characteristics and differences in sentiment expression in movie review documents from four online opinion genres – blog postings, discussion board threads, user reviews, and critic reviews.Design/methodology/approach – A collection of movie review documents was harvested from the four types of web sources, and a sample of 520 movie reviews were analysed to compare the content and textual characteristics across the four genres. The analysis focused on document and sentence length, part‐of‐speech distribution, vocabulary, aspects of movies discussed, star ratings used and multimedia content in the reviews. The study also identified frequently occurring positive and negative terms in the different genres, as well as the pattern of responses in discussion threads.Findings – Critic reviews and blog postings are longer than user reviews and discussion threads, and contain longer sentences. Critic reviews and blogs contain more nouns and prepositions, whereas discuss...