ALTAS: An Intelligent Text Analysis System Based on Knowledge Graphs

Abstract

This paper presents an intelligent text analysis system, called ALTAS, to support various text analysis tasks such as statistics analysis, sentiment analysis, text classification, and text clustering. The system contains four main components: knowledge graphs, text processing, text analysis and intelligent report. First, the system has built a semantic-rich knowledge base using several knowledge graph resources. A novel text processing and analysis framework based on knowledge graphs is developed and implemented. Given a text dataset, the text processing phase will do data cleaning, word segmentation and feature extraction for it. With the extracted features, the text analysis phase allows users to select a text mining task. We have implemented the proposed novel algorithm and several typical algorithms for each task. If users select multiple algorithms for the task, the intelligent report phase will automatically generate comparison results for users. Especially, the intelligent report phase also provides users a paper summary generating function on text mining problems.