Makoto Koyama

Learn More
Although Pseudo-Relevance Feedback (PRF) is a widely used technique for enhancing average retrieval performance, it may actually hurt performance for around one-third of a given set of topics. To enhance the reliability of PRF, Flexible PRF has been proposed, which adjusts the number of pseudo-relevant documents and/or the number of expansion terms for each(More)
Although research in effective Question Answering (QA) has become active in recent years, it was not clear how system effectiveness affects user satisfaction in a practical QA environment. We therefore considered two practical environments in which QA may be useful (namely, Desktop and Mobile) and conducted a questionnaire survey for each environment. The(More)
This paper introduces a Japanese question answering system called ASKMi (Answer Seeker/Knowledge Miner). We describe its system components with an emphasis on the Question Analyzer, which performs Sematic Role Analysis for answer type recognition, query expansion and document constraint generation. Semantic Role Analysis is a rule-based,(More)
Toshiba participated in the Japanese monolingual track and the English-Japanese cross-language track at NTCIR-3, and achieved the highest retrieval performances among the DESCRIPTION runs. Our pre-submission experiments using the KIDS system fo-cused primarily on effective Japanese monolingual retrieval, and addressed several problems including term(More)
Toshiba participated in the Monolingual/Bilingual tasks at NTCIR-4 CLIR using our CLIR system called BRIDJE. We submitted 24 runs covering three topic languages (Japanese, English and Chinese) and two document languages (Japanese and English) and achieved the highest performances in the E-J-D, C-J-D, C-J-T, E-E-D, J-E-D, J-E-T subtasks. We had 12 more runs(More)
Toshiba participated in NTCIR-4 QAC2 Subtask 1: this is our first QAC participation. Using our newly developed Japanese QA system called ASKMi, we submitted two runs, one using Okapi/BM25 for document retrieval (TSB-A) and the other using Boolean AND constraints before applying Okapi/BM25 (TSB-B). We achieved the 5th best performance among the 17(More)
This paper describes two new features of the BRIDJE system for cross-language information access. The first feature is the partial disambiguation function of the Bi-directional Retriever, which can be used for search request translation in cross-language IR. Its advantage over a " black-box " machine translation approach is consistent across five test(More)
Toshiba participated in the Monolingual and Bilingual IR tasks at NTCIR-5 CLIR using the BRIDJE system.erage Precision and Q-measure in addition to the standard Arithmetic Mean ones. Our analyses show that the Geometric Mean, which focusses on the " harder " topics, provides new insight into retrieval effectiveness evaluation.
At NTCIR-6 CLIR, Toshiba participated in the Monolingual and Bilingual IR tasks covering three topic languages (Japanese, English and Chinese) and one document language (Japanese). For Stage 1 (which is the usual ad hoc task using the new NTCIR-6 topics), we submitted two DESCRIPTION runs and two TITLE runs for each topic language. Our first search strategy(More)
PicA Topic is a prototype system designed for enabling the user to view topical segments of recorded TV shows selectively. By analysing closed captions and eletronic program guide texts, it performs topic segmentation and topic sentence selection, and presents a clickable table of contents to the user. Our previous work handled TV shows on travel, and(More)