Hitachi Software Engineering Company, Ltd.

Tadashi Mizunuma

Junji Yoshii

Shigeru Tago

Keisuke Yamamoto

Tsunehiko Watanabe

Tateo Nagai

Yuichi Minesaki

Yuji Morikawa

Fumitoshi Ogura

Katsunori Masaki

Necessary information can be easily extracted from a plurality of databasesin which biological substance information is stored. Data is downloaded from the plurality of databases. From the downloaded data is extracted information indicating links between two databases, a detailed description of each data, and sequence data for homology search, which together constitute an index. The thus extracted indexis distributed to a user facility, where a userconducts a search using the distributed index.

Data distribution method, data search method, and data search system

The invention provides information concerning records that cannot be retrieved on a search path among databases when the retrieval of search keys for other related databases and records is carried out in a plurality of databases in a chain-reactive manner. In the course of retrieving search keys and records in a chain-reactive manner up to a terminal database, information for identifying a group of search keys and a group of records retrieved through a series of retrieval processing is outputted in one line or one column of a table. When the record retrieved from a certain database in the chain-reactive search procedure does not contain a search key to be entered into a subsequent database, information for identifying a record that does not contain the search key and information for identifying a database having the record is outputted.

Database search information output method

In the preparation of a characteristic word list as one method of text mining, means for selecting an unnecessary word group that requires to be removed in advance is provided. Document IDs and a user ID are received from a client, and a characteristic word list is prepared from the document IDs and document information obtained in advance. Next, selected book IDs are obtained referring to the user ID and a user attribute table. On the basis of the obtained book IDs, unnecessary words are removed from the characteristic word list prepared by characteristic word list preparation program, and the characteristic word list from which the unnecessary words are removed is returned to the client as mining results.

Text mining server and program

A plurality of database search path candidates are automatically determined and provided. Databases containing information with similar characteristics are divided into groups (hereafter referred to as a 'category group'), and a correlation diagram is created based on links between those category groups. In the diagram, high priority is given to links between category groups constituting a path that has to be passed through due to relevance between the characteristic of databases, and normal priority is given to other links. When a path search is carried out in this diagram, an important path is selected regardless of attributes such as distance between databases or time by carrying out a first path search along the high priority path between category groups.

Growing community of inventors

Tokyo, Japan

Tadashi Mizunuma