A method for generating a machine translation (MT) dictionary from parallel texts is described. This method utilizes both statistical information and linguistic information to obtain corresponding words or phrases in parallel texts. By combining these two types of information, translation pairs which cannot be obtained by a linguistic-based method can be extntcted. Over 70% accurate translations of compound nouns and over 50% of unknown words are obtained as tbe first candidate from small Japanese/Englisb parallel texts containing severe distortions.

In this paper we report results of an investigation into EnglishJapanese Cross-Language Information Retrieval (CLIR) comparing a number of query translation methods. Results from experiments using the standard BMIR-J2 Japanese collection suggest that full machine translation (MT) can outperform popular dictionary-based query translation methods and further that in this context MT is largely robust to queries with little linguistic structure.

No abstract

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 collections and across two language permutations: English-Japanese and Japanese-English. The second new feature is the Information Distiller, which performs interactive summarisation of retrieved documents based on Semantic Role Analysis. Our examples illustrate the usefulness of this feature, and our evaluation results show that the precision of Semantic Role Analysis is very high.

We have developed an optimal scheduling method for raw material operations aiming the raw materials cost reduction. In this paper, we report optimization approaches to minimize the cost of ore blending in steel works. The ore blending problem is to make schedules for the purpose of cost minimization under several constraints such as the stock in yards, ingredients in sintered ore. When formulating as a mathematical model, nonlinearity exists in this problem and make it complicated. However, this problem has characteristic that becomes a linear problem by fixing several key variables as constants. To overcome the nonlinearity, we developed our original Hybrid model that was a combination of Particle Swarm Optimization (PSO) and Linear Programming method (LP). We applied PSO to search the best way of fixing key variables, and obtained blending schedules by solving LPs. Our Hybrid model searched wide area effectively, and derived the solution within 2 minutes. Numerical experiments indicated a cost reduction of secondary materials by 1%.