Hyopil Shin

Schema and constraints-based matching and merging of Topic Maps

In this paper, we propose a multi-strategic matching and merging approach to find correspondences between ontologies based on the syntactic or semantic characteristics and constraints of the Topic Maps. Our multi-strategic matching approach consists of a linguistic module and a Topic Map constraints-based module. A linguistic module computes similarities between concepts using morphological analysis, string normalization and tokenization and language-dependent heuristics. A Topic Map constraints-based module takes advantage of several Topic Maps-dependent techniques such as a topic property-based matching, a hierarchy-based matching, and an association-based matching. This is a composite matching procedure and need not generate a cross-pair of all topics from the ontologies because unmatched pairs of topics can be removed by characteristics and constraints of the Topic Maps. Merging between Topic Maps follows the matching operations. We set up the MERGE function to integrate two Topic Maps into a new Topic Map, which satisfies such merge requirements as entity preservation, property preservation, relation preservation, and conflict resolution. For our experiments, we used oriental philosophy ontologies, western philosophy ontologies, Yahoo western philosophy dictionary, and Wikipedia philosophy ontology as input ontologies. Our experiments show that the automatically generated matching results conform to the outputs generated manually by domain experts and can be of great benefit to the following merging operations.

Tag Sense Disambiguation for Clarifying the Vocabulary of Social Tags

Tagging is one of the most popular services in Web 2.0. As a special form of tagging, social tagging is done collaboratively by many users, which forms a so-called folksonomy. As tagging has become widespread on the Web, the tag vocabulary is now very informal, uncontrolled, and personalized. For this reason, many tags are unfamiliar and ambiguous to users so that they fail to understand the meaning of each tag. In this paper, we propose a tag sense disambiguating method, called Tag Sense Disambigu-ation (TSD), which works in the social tagging environment. TSD can be applied to the vocabulary of social tags, thereby enabling users to understand the meaning of each tag through Wikipedia. To find the correct mappings from del.icio.us tags to Wikipedia articles, we define the Local Neighbor tags, the Global Neighbor tags, and finally the Neighbor tags that would be the useful key-words for disambiguating the sense of each tag based on the tag co-occurrences. The automatically built mappings are reasonable in most cases. The experiment shows that TSD can find the cor-rect mappings with high accuracy.

KTimeML: Specification of Temporal and Event Expressions in Korean Text

TimeML, TimeBank, and TTK (TARSQI Project) have been playing an important role in enhancement of IE, QA, and other NLP applications. TimeML is a specification language for events and temporal expressions in text. This paper presents the problems and solutions for porting TimeML to Korean as a part of the Korean TARSQI Project. We also introduce the KTTK which is an automatic markup tool of temporal and event-denoting expressions in Korean text.

FolksoViz: A Semantic Relation-Based Folksonomy Visualization Using the Wikipedia Corpus

Tagging is one of the most popular services in Web 2.0 and folksonomy is a representation of collaborative tagging. Tag cloud has been the one and only visualization of the folksonomy. The tag cloud, however, provides no information about the relations between tags. In this paper, targeting del.icio.us tag data, we propose a technique, Folk-soViz, for automatically deriving semantic relations between tags and for visualizing the tags and their relations. In order to find the equivalence, subsumption, and similarity relations, we apply various rules and models based on the Wikipedia corpus. The derived relations are visualized ef-fectively. The experiment shows that the FolksoViz manag-es to find the correct semantic relations with high accuracy.

Automatic Product Review Helpfulness Estimation based on Review Information Types

온라인 상품평 양의 비약적 증가로 인해 소비자들이 유용한 상품평 만을 찾는 것이 거의 불가능에 가까워졌다. 이 연구는 온라인 상품평의 유용성을 자동적으로 평가할 수 있는 토대를 마련하는데 그 목적이 있다. 이를 위해 상품평을 이루는 문장에 담긴 정보를 설명하는 그 대상에 따라 종류를 나눌 수 있도록 상품평 정보 분류를(Review Information Types) 제안하고, 각 정보 분류 내에서 문장의 주제 벡터 변환 방법과 군집화를 이용하여 더 세부적으로 각 문장이 어떤 정보를 제공하는지를 추출함으로써 각 상품평이 제공하는 정보에 따라 그 유용성을 평가하는 방법을 제안한다. 이러한 시도는 잠재적 소비자들이 상품평에서 상품 자체의 특성이나 상품평 제공자의 경험과 같은 정보를 배송과 같은 정보보다 중요하게 생각할 것이라는 가정에서 시작했다. 자동 상품평 유용성 평가 실험을 통해 본 연구에서 제시하는 방법이 기존의 비교 가능한 연구들에 비해 더 효과적인 것을 밝혀냈다.

Do We Really Need Lexical Information? Towards a Top-down Approach to Sentiment Analysis of Product Reviews

Most of the current approaches to sentiment analysis of product reviews are dependent on lexical sentiment information and proceed in a bottom-up way, adding new layers of features to lexical data. In this paper, we maintain that a typical product review is not a bag of sentiments, but a narrative with an underlying structure and reoccurring patterns, which allows us to predict its sentiments knowing only its general polarity and discourse cues that occur in it. We hypothesize that knowing only the review’s score and its discourse patterns would allow us to accurately predict the sentiments of its individual sentences. The experiments we conducted prove this hypothesis and show a substantial improvement over the lexical baseline.

Pinpointing Sentence-Level Subjectivity through Balanced Subjective and Objective Features

The sentence-level subjectivity classification is a challenging task. This paper pinpoints some of its unique characteristics. It argues that these characteristics should be considered when extracting subjective or objective features from sentences. Through various sentence-level subjectivity classification experiments with numerous feature combinations, we found that balanced features for both subjective and objective sentences help to achieve balanced precision and recall for sentence subjectivity classification.

Problematic Set-Denoting Temporal Expressions in the Framework of ISO-TimeML

This study investigates how to represent set-denoting temporal expressions with ISO-Time ML, a state-of-the-art framework for representing time expressions. Specifically, we will show its limitations, caused by characteristics of TIMEX3, i) the freq attribute can not represent both the frequency of time points and the time granularity at the same time. ii) set expressions structurally containing another set expressions can not be annotated in the current framework. iii) value attribute can not contain values of complex set-denoting expressions. iv) there are set-expressions that have no set marker. These limitations reveal that TIMEX3 needs to be modified or extended to represent those problematic temporal expressions.

A Noun-Predicate Bigram-Based Similarity Measure for Lexical Relations

The method outlined in this paper demonstrates that the information-theoretic similarity measure and noun-predicate bigrams are effective methods for creating lists of semantically-related words for lexical database work. Our experiments revealed that instead of serious syntactic analysis, bigrams and morpho-syntactic information sufficed for the feature-based similarity measure. We contend that our method would be even more appreciated if it applied to a raw newswire corpus in which unlisted words in existing dictionaries, such as recently-created words, proper nouns, and syllabic abbreviations, are prevailing.