Research On The Method Of Project Knowledge Context Extracting Based On Topic Models

Project knowledge context (PKC) are descriptions of situation and background information generated during the creation, acquiring, delivering, sharing and applying of project knowledge in project activities or processes, as well as a critical component of project knowledge. A deep known about project knowledge context can be an advantage in better understanding of knowledge. As project knowledge context hiding in literal terms of project data, it is a good approach to model, to extract and to store project knowledge context from literal project data while doing research on issues about project management and project knowledge.This paper is a research about topic based method of PKC extraction from project data. It discussed relationships between PKC and topics of literal project documents, revealed potential topics from documents, analyzed contextual factors presented by document topics. Considering the contextual factors in project knowledge retrieval, recommendation and other applications, this paper gives a result that project knowledge context can obvious improve efficiency in project knowledge management, sharing and learning.This paper launched literature analysis first. The literature analysis provided discussion of research progress of PKC. Then conducted by research and design methodology, this paper explored possible approaches to build PKC models. Combining with the topic models, this paper innovative designed topic model based PKC model, method of similarity measurement of PKC and the validating process PKC model.The main research work of this paper can be divided into the following three parts:First, based on the literature review of PCK and topic model, this paper discussed the relationships between document topics and PKC. the discussion provides an instruction about how to build a method of extracting PKC based on topic models. Following the the discussion and the instruction, this paper gained an approach to build a multi-dimensional and multi-level PKC model, PKC extraction method and algorithms accordingly.Then this paper built a prototype PKC extracting system based on these algorithms, PKC model and PKC extracting method. This prototype system implemented all the algorithms and concatenated all the sophisticated steps into a data flowing process.Last, this paper launched an evaluation framework to validate the method of PKC extracting based on topic models and prototype system. The comparision of this method and two other information intrieval and recommendation methods based on the project document generated by Beijing-Shanghai high-speed railway, the results shows that the method of PKC extracting based on topic model has a prominent advantage in the accuracy of knowledge retrieval.In summary, this paper, based on the research and design methodology, completed the practice research of PKC extracting based on topic models with a serial steps of theoretical demonstration, model design, algorithm design, prototype system construction and evaluation, explored a method combined of natural language processing and PKC acquisition in knowledge management. This study provides a theoretical method and practical basis to improve the information and knowledge level and to increase knowledge sharing in organizations and enterprises.