Research On Domain Knowledge Base Construction Platform For Selective Dissemination Of Information

The amount of literature resources under the environment is huge, and it update rapidly, it can’t meet the need of library or information institutations only relys on the traditional way of literature processing and handling, so this article proposed the research on domain knowledge base construction platform for selective dissemination of information. In this article, we mainly study the service mode, logical architecture, technical system and application issues, discusses the key technologies involved in the platform from the perspective of system development and application. On this basis, we design and build the domain knowledge base construction platform for selective dissemination of information. The platform can assist intelligence officers to build domain knowledge base, thus further improving the efficiency of literature resources processing, handling, knowledge organization and utilization and the capacity of service.The main work and research of this article includes the following four aspects:(1) For the problem of a variety of heterogeneous literature resource collection, organization and processing, combined with the characteristics of literature and needs of selective dissemination of information, we achieve the function of web literature resources automatic collection. On this basis, we proposed the uniform resource description framework for selective dissemination of information reference to the Dublin Core model, it supports common literature resources description and storage, and it facilitates unified management of heterogeneous resources, cross-platform resource sharing and further processing and utilization.(2) Based on the existing algorithms and models, we proposed the improved algorithms of keywords extraction, summarization and text classification. Based on the statistical model and N-gram language model, combined with the frequency, length, part of speech, location and other information of characteristics, we proposed the algorithm of keyword extraction for the literature resource of selective dissemination of information, which is a field-independent algorithm and can identify unlisted word effectively. Based on the statistics based summarization algorithms, combined with the content and location information of sentence, to evaluate the weight of candidate summary sentence, we proposed the algorithm of automatic summary extraction, which can filter redundant summary sentence by calculate the similarity between the candidate summary sentences. Based on the Naive Bayes-based classification algorithm, we achieve the automatic text classification of literature resources, which setting different weights for the characters of training data and testing data, it achieved better classification results than Naive Bayes classification algorithm.(3) From the perspective of macro analysis of literature resources, analyse the topics and external or internal features using HDP topic model, based on the HDP topic model, we proposed the literature recommended model for literature retrieval, which can convert the similarity calculation from literature characters to topics implicated the semantic of characters and improve the accuracy of literature recommendation. In addition, the literature recommended model also introduces the time threshold, which can recommend literatures closer to the current literature or published lately in literature retrieval.(4) Based on the.Net platform, we design and implement the domain knowledge base platform for selective dissemination of information. The platform can achieve literature resources collection, organization, processing, utilization and many other functions. Through relevant experiments, it has been proved the technical feasibility of this platform, and provide some reference for relative research and application.