New Approach To Specification Of Correct Requests For Data Mining
Free (open access)
O Ryabov & K Mori
The project addresses the problem of \“unskilled” human request expression and assimilation. The same phenomenon, the same object may be described in different ways by different people. Further, different recipients can interpret the same sentence differently. Orthographic or grammatical errors make the problem even worse. This is the main barrier for development of any system for recognizing the meaning of human speech, text, or information search request in particular. As the result, the most of current data base interfaces leave the step of the request definition to the users. This drastically limits the data mining effectiveness for them. The originality of the proposed approach is based on designing a request specification interface with direct assistance functionality. It helps client to describe his request absolutely correctly being sure the system \“understands” him completely. As the systems reaction the client may expect fruitful data mining results every time. 1 Introduction In this paper we would like to focus our study on the problem of \“unskilled” user access to the data base systems. The user interface is a tool for interpretation of the user information or data search request into the data mining query. The \“unskilled” user is a person having neither information search experience, nor special knowledge on the data he looks for. Needless to say, the user interface is a key part of any data base system, especially in the cases assuming the general public access. The entire system quality, as well as its usefulness, is evaluated by the quality of the interface.