The demand for a tool for summarizing emerging topics is increasing in modern life since the tool can deliver well-organized information to its users. Even though there are already a number of successful search systems, the system which automatically summarizes and organizes the content of emerging topics is still in its infancy. To fulfill such demand, we introduce an automated report generation system that generates a well-summarized human-readable report for emerging topics. In this report generation system, emerging topics are automatically discovered by a topic model and news articles are indexed by the discovered topics. Then, a topical summary and a timeline summary for each topic is generated by a topical multi-document summarizer and a timeline summarizer respectively. In order to enhance the apprehensibility of the users, the proposed report system provides two report modes. One is Today's Briefing which summarizes five discovered topics of every day, and the other is Full Report which shows a long-term view of each topic with a detailed topical summary and an important event timeline.

Knowledge base completion aims to complete a knowledge base by filling up missing facts of the knowledge base. Neural knowledge base embeddings proposed to solve this task measure the plausibility of all candidate triples, and then select top-ranked triples by the plausibility as new facts for the knowledge base. The plausibility by neural embeddings allows true facts to be ranked at high positions, but not at top positions. This is because neural knowledge base embeddings are limited to using only the information within the knowledge base. Therefore, this paper proposes a re-ranking model for precise knowledge base completion. As a re-ranking model, a neural network which uses knowledge base schema and web statistic additionally is adopted. As a result, the proposed re-ranking model has an effect of using additional information for knowledge base completion. Thus, the candidate triples are first ranked by a neural knowledge base embedding, and then the result is re-ranked by the neural network. The experimental results show that the proposed re-ranking model improves the base neural embeddings up to 16% in Hits@1. This implies that the re-ranking model places true facts at top positions effectively.

Hee-Geun Yoon, Hyun-Je Song, Seong-Bae Park, Se-Young Park. Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2016.

This paper proposes a topic-based method to reflect calls and proximities simultaneously into finding interaction patterns from a mobile log. For this purpose, the proposed method regards calls and proximities as a homogeneous information type that are drawn from the same temporal space expressed by the same distribution, but with different parameters. The number of proximities in a mobile log usually overwhelms that of calls and the proximities are observed regularly. Therefore, the proposed method models a single directional influence from proximities to calls, where both call and proximity are modeled by the Latent Dirichlet Allocation (LDA). According to the experiments on the data set from MIT's Reality Mining project, the proposed method outperforms the method that treats calls and proximities independently, which proves the plausibility of the proposed method.

One of the critical problems in natural language interfaces is the discordance between the expressions covered by the interface and those by the knowledge base. In the graph-based knowledge base such as an ontology, all possible queries can be prepared in advance. As a solution of the discordance problem in natural language interfaces, this paper proposes a method that translates a natural language query into a formal language query such as SPARQL. In this paper, a user query is translated into a formal language by choosing the most appropriate query from the prepared queries. The experimental results show a high accuracy and coverage for the given knowledge base.

This paper proposes a novel method to translate tags attached to multimedia contents for cross-language retrieval. The main issue in this problem is the sense disambiguation of tags given with few textual contexts. In order to solve this problem, the proposed method represents both tags and its translation candidates as networks of co-occurring tags since a network allows richer expression of contexts than other expressions such as co-occurrence vectors. The method translates a tag by selecting the optimal one from possible candidates based on a network similarity even when neither the textual contexts nor sophisticated language resources are available. The experiments on the MIR Flickr-2008 test set show that the proposed method achieves 90.44% accuracy in translating tags from English into German, which is significantly higher than the baseline methods of a frequency based translation and a co-occurrence-based translation.

Information extraction from world wide web has been paid great attention to. Since a table is a well-organized and summarized knowledge expression for a domain, it is of great importance to extract information from the tables. However, many tables in web pages are used not to transfer information but to decorate the pages. Therefore, it is one of the most critical tasks in web table mining to discriminate the meaningful tables from the decorative ones. The main obstacle of this task comes from the difficulty of generating relevant features for the discrimination. This paper proposes a novel method to discriminate them using a composite kernel which combines a parse tree kernel and a linear kernel. Since a web table is represented as a parse tree by a HTML parser, the parse tree kernel can be naturally used in determining the similarity between trees, and the linear kernel with content features is used to make up for the weak points of the parse tree kernel. The support vector machines with the composite kernel distinguish with high accuracy the meaningful tables from the decorative ones. A series of experiments show that the proposed method achieves the state-of-the-art performance.

The ontology alignment has two kinds of major problems. First, the features used for ontology alignment are usually defined by experts, but it is highly possible for some critical features to be excluded from the feature set. Second, the semantic and the structural similarities are usually computed independently, and then they are combined in an ad-hoc way where the weights are determined heuristically. This paper proposes the modified parse tree kernel (MPTK) for ontology alignment. In order to compute the similarity between entities in the ontologies, a tree is adopted as a representation of an ontology. After transforming an ontology into a set of trees, their similarity is computed using MPTK without explicit enumeration of features. In computing the similarity between trees,the approximate string matching is adopted to naturally reflect not only the structural information but also the semantic information. According to a series of experiments with a standard data set, the kernel method outperforms other structural similarities such as GMO. In addition, the proposed method shows the state-of-the-art performance in the ontology alignment.

An automatic word spacing is one of the important tasks in Korean language processing and information retrieval. Since there are a number of confusing cases in word spacing of Korean, there are some mistakes in many texts including news articles. This paper presents a high-accurate method for automatic word spacing based on self-organizing η-gram model. This method is basically a variant of η-gram model, but achieves high accuracy by automatically adapting context size.In order to find the optimal context size, the proposed method automatically increases the context size when the contextual distribution after increasing it dose not agree with that of the current context. It also decreases the context size when the distribution of reduced context is similar to that of the current context. This approach achieves high accuracy by considering higher dimensional data in case of necessity, and the increased computational cost are compensated by the reduced context size. The experimental results show that the self-organizing structure of η-gram model enhances the basic model.

An easy way of translating queries in one language to the other for cross-language information retrieval (IR) is to use a simple bilingual dictionary. Because of the general-purpose nature of such dictionaries, however, this simple method yields a severe translation ambiguity problem. This paper describes the degree to which this problem arises in Korean-English cross-language IR and suggests a relatively simple yet effective method for disambiguation using mutual information statistics obtained only from the target document collection. In this method, mutual information is used not only to select the best candidate but also to assign a weight to query terms in the target language. Our experimental results based on the TREC-6 collection shows that this method can achieve up to 85% of the monolingual retrieval case and 96% of the manual disambiguation case.

This report introduces the current situation of machine translation in Korea. Recently, the need for further developing machine translation has been generally recognized. Although a few machine translation softwares for Korean have been released on the market, they do not sufficiently meet the need of users. As a result, the Korean machine translation field is only a niche market. However, several projects are underway in Korea which include world-wide technical cooperation. This report surveys the history of machine translation in Korea and describes the current market, R&D status, and current technical difficulties.

Presents a new approach for word sense disambiguation, specially for Korean transitive verbs, based on a supervised method. This approach integrates a set of clues based on syntactic relations for word sense disambiguation. Instead of using an unordered set of the surrounding words of a given ambiguous verb, we use co-occurrences as a clue for disambiguation based on some syntactic relations such as predicate-argument, modified-modified, adjacency, or a mixture of these. The experimental results show that the accuracy of our disambiguation method on highly ambiguous verbs with fine-grained sense distinctions is always better than the default strategy of assigning all occurrences to the most frequent sense.