We deal with Chinese, Japanese and Korean multilingual information retrieval (MLIR) in NTCIR-6, and submit our results on the C-CJK-T and C-CJK-D subtask. In these runs, we adopt Dictionary-Based Approach to translate query terms. In addition to tradition dictionary, we incorporate the Wikipedia as a live dictionary.

This report is an overview of Cross-Language Information Retrieval Task (CLIR) at the third NTCIR Workshop. There are 3 tracks in CLIR: Single Language IR (SLIR), Bilingual CLIR (BLIR), and Multilingual CLIR (MLIR). The scope, schedule, test collections, search results, relevance judgment, scoring results, and the preliminary analyses are described in the report.

Sparse tensors arise in problems science, engineering, machine learning, and data analytics. Programs that operate on such can exploit sparsity to reduce storage requirements computational time. Developing maintaining sparse software by hand, however, is a complex error-prone task. Therefore, we propose treating as property of tensors, not tedious implementation task, letting compiler generate ...

Abstract Machine learning compilers are large software containing complex transformations for deep models, and any buggy transformation may cause a crash or silently bring regression to the prediction accuracy performance. This paper proposes an SMT-based translation validation framework Multi-Level IR (MLIR), compiler used by many compilers. It SMT encoding tailored that is over-approximation ...

National Taiwan University (NTU) Natural Language Processing Laboratory (NLPL) participated in MLIR task in CLEF 2002. We submitted five official multilingual runs. In this paper, we try to resolve the collection fusion problem. We experimented with several merging strategies that merge the results of several intermediate runs.

This paper describes our retrieval system for NTCIR-2 Japanese/English CLIR and MLIR tasks. We integrate query and document translation with monolingual retrieval to improve retrieval accuracy, and perform clustering to improve browsing efficiency. We also introduce an entropy-driven technique in evaluating clustering methods.

The performance evaluation of information retrieval systems has achieved a high momentum in the last few years. Basic performance measures of information retrieval systems include precision and recall. While these measures work well in monolingual web retrieval, they are not suitable for CLIR (Cross-lingual Information Retrieval) and MLIR (Multilingual Information Retrieval) where two or more l...

BACKGROUND Low engraftment and adverse immune reactions hamper the success rate of clinical islet transplantation. In this study, we investigated the capacity of human mesenchymal stem cells (MSCs) to adhere to human islets of Langerhans and their effects in immune modulation and during blood interactions in vitro. METHODS Composite MSC-islets were formed by suspension co-culture, and the phe...

We participated in the SLIR, BLIR(PLIR) and MLIR subtasks of the NTCIR-5 CLIR task. Our IR system uses language models for document scoring and query expansion, and can handle four languages; Chinese, Japanese, Korean and English. The system utilizes multiple language resources (bilingual dictionaries, parallel corpora and machine translation systems). We attempted to use some techniques includ...

Habitual switching of languages is a common behaviour among polyglots when searching for information on the Web. Studies in retrieval (IR) and multilingual (MLIR) suggest that part reason such regular topic search. Unlike survey-based studies, this study uses query click-through logs. It exploits querying results selection Swahili MLIR system users to explore how search (query) associated with ...