A Decade after TREC-4 - NTCIR-5 CLIR-J-J Experiments at Yahoo!Japan
نویسنده
چکیده
This paper describes NTCIR-5 experiments of the CLIR-J-J task, i.e. Japanese monolingual retrieval subtask, at the Yahoo group, focusing on comparative studies of the feedback effectiveness with two retrieval methods, namely BM25TF*IDF and a KL-divergence language modeling approaches. An “automatic feedback from top k documents” strategy was surprisingly successful in this test collection. We compared behaviors of the systems with past NTCIR and TREC experiments and find out the characteristics of test collections where the strategy is especially effective.
منابع مشابه
NTCIR-6 CLIR-J-J Experiments at Yahoo! Japan
This paper describes NTCIR-6 experiments of the CLIRJ-J task, i.e. Japanese monolingual retrieval subtask, at the Yahoo group, focusing on the parameter optimization in information retrieval (IR). Unlike regression approaches, we optimized parameters completely independent from retrieval models so that the optimized parameter set can illustrate the characteristics of the target test collections...
متن کاملRevisiting Document Length Hypotheses: NTCIR-4 CLIR and Patent Experiments at Patolis
NTCIR-4 experiments of CLIR J-J and Patent tasks, focusing on comparative studies of two testcollections and two retrieval approaches in view of document length hypotheses are described. TF*IDF outperformed the language modeling approach in the CLIR J-J task while two approaches performed similarly in the Patent task. Two different document length hypotheses behind two tasks/collections are ass...
متن کاملNTCIR-3 CLIR Experiments at MSRA
This paper describes three statistical models for the purpose of resolving query translation ambiguity for cross-language information retrieval (CLIR). First, a decaying co-occurrence model is present. It is an extension of traditional co-occurrence models in that it contains a decaying factor which decreases the mutual information when the distance between the terms increases. Second, a phrase...
متن کاملNTCIR-4 CLIR Experiments at Oki
We participated in SLIR, BLIR(PLIR) and MLIR subtasks at the NTCIR-4 CLIR task. Our IR system can handle queries and documents in Chinese, English and Japanese. The system utilizes multiple language resources (bilingual dictionaries, parallel corpora and machine translation systems) for query translation. We adopted the pivot language approach for C-J and J-C search using English as a pivot lan...
متن کاملNTCIR-6 CLIR Experiments at Osaka Kyoiku University - Term Expansion Using Online Dictionaries and Weighting Score by Term Variety
This paper describes experimental results of J-J subtask of NTCIR-6 CLIR. We expanded query term using online dictionaries in a WEB. It was effective for some topics of which average precision was low. Probabilistic model were employed for scoring, and we modified this score multiplying by the number of varieties of query terms, also. In most cases this works well. Query term reduction should b...
متن کامل