Yohei Seki

Multilingual opinion holder identification using author and authority viewpoints

Opinion holder identification research is important for discriminating between opinions that are viewed from different perspectives. We propose a new opinion holder identification method that is based on a differentiation between the author and authority viewpoints in opinionated sentences. In our method, the author- and authority-opinionated sentences were extracted, respectively, by utilizing the different features because their writing styles were different. Although the researchers have not focused on it, this differentiation is important for correctly identifying opinion holders. We describe our participation in the NTCIR-6 Opinion Analysis Pilot Task by focusing on the opinion holder identification results in Japanese and English. The evaluation results showed that our system performed fairly well with respect to Japanese documents, and postsubmission analysis has revealed that improvements could be made with respect to English documents as well.

Opinion holder extraction from author and authority viewpoints

Opinion holder extraction research is important for discriminating between opinions that are viewed from different perspectives. In this paper, we describe our experience of participation in the NTCIR-6 Opinion Analysis Pilot Task by focusing on opinion holder extraction results in Japanese and English. Our approach to opinion holder extraction was based on the discrimination between author and authority viewpoints in opinionated sentences, and the evaluation results were fair with respect to the Japanese documents.

Multi-Document Viewpoint Summarization Focused on Facts, Opinion and Knowledge

An interactive information retrieval system that provides different types of summaries of retrieved documents according to each user’s information needs, situation, or purpose of search can be effective for understanding document content. The purpose of this study is to build a multi-document summarizer, “Viewpoint Summarizer With Interactive clustering on Multidocuments (v-SWIM)”, which produces summaries according to such viewpoints. We tested its effectiveness on a new test collection, ViewSumm30, which contains human-made reference summaries of three different summary types for each of the 30 document sets. Once a set of documents on a topic (e.g., documents retrieved by a search engine) is provided to v-SWIM, it returns a list of topics discussed in the given document set, so that the user can select a topic or topics of interest as well as the summary type, such as fact-reporting, opinion-oriented or knowledge-focused, and produces a summary from the viewpoints of the topics and summary type selected by the user. We assume that sentence types and document genres are related to the types of information included in the source documents and are useful for selecting appropriate information for each of the summary types. “Sentence type” defines the type of information in a sentence. “Document genre” defines the type of information in a document. The results of the experiments showed that the proposed system using automatically identified sentence types and document genres of the source documents improved the coverage of the system-produced fact-reporting, opinion-oriented, and knowledge-focused summaries, 13.14%, 34.23%, and 15.89%, respectively, compared with our baseline system which did not differentiate sentence types or document genres.

Compact Summarization for Mobile Phones

In this paper, we propose a new summarization method appropriate for sending text to mobile phones. In mobile access research, an important issue is how to display compact and informative summaries on a screen much smaller than that of an ordinary computer. Documents with varieties of genres presenting information such as opinions, evaluations, etc. have been published on the Web. Most previous summarization research, however, has focused on factual information and topics in documents. For a document that asserts the author’s opinion, we assumed that combining factual information and subjective information such as opinions would be effective to produce short but informative summaries adequate to comprehend the contents of the original documents. We propose a summarization method that exploits the typical text structure of the genre. We test the effectiveness of the proposed methods by asking three users who use the genre of ”columns” in ordinary life to evaluate summaries in aspect of the recognition test of important sentences and to demonstrate their comprehension of original documents. With the comprehension test, our method which was based on the usage of sentence types was evaluated to be more informative than the existing methods.

Automatic alignment of ontology eliminating the probable misalignments

This paper describes a novel approach of detecting misalignment at the time of aligning two different ontologies, and of eliminating the misalignments. Our objective is to reduce limitation of a specific technique of ontology alignment. Two aligned sets extracted by different alignment techniques from the same pair of ontology, are fed to the misalignment detection and elimination process to produce better alignments. Our experiments demonstrate that our method, taking advantage of misalignment detection and elimination, shows a good recall and precision.

Use of Twitter for Analysis of Public Sentiment for Improvement of Local Government Service

Active collaboration with the public is the key to improving the administrative services of local government, and social media is an essential tool for understanding public sentiment. In this paper, we propose a method to gauge public sentiment for local government by using Twitter. We conducted an operational test with local government officers and found that our proposal was effective in revealing sentiments that were meaningful for the public, such as avoiding full parking lots, finding potholes in the road, or attracting favorable comments or suggestions to improve festival events.

Finding impressive social content creators: searching for SNS illustrators using feedback on motifs and impressions

We propose a method for finding impressive creators in online social network sites (SNSs). Many users are actively engaged in publishing their own works, sharing visual content on sites such as YouTube or Flickr. In this paper, we focus on the Japanese illustration-sharing SNS, Pixiv. We implement an illustrator search system based on user impression categories. The impressions of illustrators are estimated from clues in the crowdsourced social-tag annotations on their illustrations. We evaluated our system in terms of normalized discounted cumulative gain and found that using feedback on motifs and impressions for illustrations of relevant illustrators improved illustrator search by 11%.

Twitter User Classification with Posting Locations

Twitter contains a large number of postings related to the reputation of products and services. Analyzing these data can provide useful marketing information. Inferring the user class would make it possible to extract opinions related to each class. In this paper, we propose a method that treats each user’s posting location for a tweet as a feature in the analysis of user classes. The proposed method creates clusters of geotags (obtained from Twitter tags) to identify the locations most often visited by the target user, which are then used as features. As an example, we conducted experiments to classify targets based on three classes: “student,” “working member of society,” and “housewife.” We obtained an average F-measure of 0.779, which represents an improvement on baseline results.

2nd Asian Summer School in Information Access (ASSIA 2015)

The first Asian Summer School in Information Access (ASSIA 2013) was held between 22nd and 24th June, 2013 in Tsukuba, Japan. The summer school offered 9 lectures in Information Retrieval, Web Search, and related topics, along with two panel discussions and a poster session. This reports a successful international summer school in Asia attracting a total of 63 participants from the range of countries in Asia, Europe, and North America.