Method and system for discriminating and eliminating false positive results

Abstract

The invention discloses a method and a system for discriminating and eliminating false positive results, so that the false positive rate in a literature database knowledge mining process can be efficiently reduced. The technical scheme is that the method comprises: performing optimization processing on a target literature database and a controlled vocabulary; performing classification on false positive results and establishing a discovery mode; generating a false positive result set; extracting the false positive results and original data from the false positive result set, and performing classification; performing grouping on the false positive results of all types, extracting data in mining results of all types, performing training on the data by using an algorithm library to obtain a proper elimination algorithm, performing checking by using residual grouping data, and if the checking is passed, selecting the elimination algorithm, otherwise, modifying the algorithm library and performing re-training until the proper elimination algorithm is found; based on the found elimination algorithm, constructing a false positive elimination dictionary and a corresponding false positive elimination logic algorithm library; and traversing the whole false positive result set, eliminating all the false positive results, feeding back information of eliminating the false positive results to the target literature database, and finally correcting all false positive mining results in the target literature database.

Description

technical field [0001] The invention relates to the technology in the knowledge mining of the document database, in particular to a method and a system for screening and eliminating false positive results in the knowledge mining of the document database. Background technique [0002] Document knowledge mining refers to the process of extracting, integrating and discovering useful information and knowledge points from documents. Through document knowledge mining, a large number of documents can be quickly processed and knowledge in a specific field can be obtained. Document knowledge mining involves data mining, text mining, natural language Processing and information integration and other research areas. Taking life science literature knowledge mining as an example to illustrate, the main content of life science literature knowledge mining research is divided into five parts: information retrieval, entity recognition, information extraction, text mining and information integ...

AI Technical Summary

Problems solved by technology

However, in the process of knowledge mining and discovery of databases, false positive knowledge mining results are unavoidable. The reasons for this phenomenon mainly come from the following three aspects: ①The quality of raw data collection in the database is low; ②In the text mining tool dictionary In the process of compiling, the accuracy and uniformity of the usage of professional words is not enough; ③In the process of information integration and knowledge mining, the selection of computer algorithms and mining methods and approaches is unreasonable or the algorithm itself has systematic defects

[0004] At present, although the global research on database knowledge mining and discovery is at a relatively hot stage, the research on database knowledge mining false positives is still in the exploratory stage. The relevant patents and published publications in this research field There are not many related research reports

Moreover, a comprehensive method for eliminating false positive mining results in the process of document knowledge mining has not been found so far. Therefore, it is urgent to develop a new comprehensive method and approach for screening and eliminating false positive results of document knowledge mining.

Images