Knowledge graph construction-oriented text time extraction and standardization method

Abstract

The invention relates to a knowledge graph construction-oriented text time extraction and standardization method. The method comprises the following steps of constructing a time information knowledgebase; according to a time expression template in the time information knowledge base, extracting time phrases, prepositions and time nouns of locality from a to-be-identified text, and automatically mapping the time phrases into absolute time expressions in sequence; judging semantic types of the time phrases by utilizing a time semantic modeling algorithm; and outputting a time semantic model quintuple TSM=(AT, RTP, PP, PD, ST). Compared with the prior art, the method has the advantages that data, about time information, of knowledge tuples in a knowledge graph is supplemented; the problem oftime information mapping is solved; a high-quality time information base is formed; and the like.

Description

technical field [0001] The present invention relates to an information extraction method in the field of big data and natural language processing, in particular to a text time extraction and standardization method oriented to knowledge map construction. Background technique [0002] Knowledge tuples, the basic elements of a knowledge map, can be divided into commonsense knowledge and event-based knowledge. Event-based knowledge has a strong real-time nature, and only when time information is integrated into knowledge tuples can it fully and accurately express all of them. the knowledge contained. For event-based knowledge tuples extracted from news texts, timeliness requirements are high. Accurately grasping the time context of knowledge existence and obtaining time semantics is of great significance for improving knowledge tuples and building knowledge graphs. [0003] Time exists objectively, but it needs to be described with the help of natural language. Temporal inform...

AI Technical Summary

Problems solved by technology

Among them, there are many researches on time information extraction, generally using rule-based methods or machine learning-based methods. The rule-based method is simple, but it is difficult to manually formulate an accurate and comprehensive rule system.

Model training based on machine learning must require a certain scale of labeled training data, and the cost of manual labeling is high

In addition, due to the flexible and diverse forms of time information, there are many difficulties in the mapping of Chinese time information, such as relative time conversion problems, incomplete time completion problems, etc., resulting in poor normalization of time phrases.

Images