Abstract:

This study investigates the methods of extracting, aggregating and visualizing interdisciplinary information sources based on literature reviews, and the information source map in the interdisciplinary domain of “machine learning” has been built by taking the informationsource information from highquality literature. Firstly, text preprocessing was carried out on the collected literatures; secondly, the auxiliary dictionary has been constructed on the basis of recognizing the context relationship, the word segmentation has been finished by using dictionaries and auxiliary dictionaries, and the entity recognition and part of speech tagging have been completed. According to the result of entity recognition and rules of syntax logic, sentence extraction and information source data extraction have been achieved; Moreover, information from the same information source in multiple articles has been integrated. Finally, the map of information source has been accomplished by using D3 tool from the dimensions of time and source. Experiments indicate that this method could lead to good results with high recall and precision ratio, which is helpful for interdisciplinary researchers to quickly master the developing process of the field, the high value information source and the characteristics of research of direction different disciplines.

Key words: Interdisciplinary, Information source map, Literature reviews, Information extraction; , Entity identification, Visualization