2017 3rd International Symposium on Mechatronics and Industrial Informatics (ISMII 2017)
ISBN: 978-1-60595-501-8
Named Entities Recognition in Computer Field for Entity Attribute
Semantic Knowledge Database
Honglin Wu
1,a
, Ruoyi Zhou
2,b
and Ke Wang
3,c
1
College of Computer Science and Engineering, Northeastern University, Shenyang, China
2
School of Information Engineering, Zhengzhou University, Zhengzhou, China
3
Research Center for Artificial Intelligence, Shenyang Linge Technology Co., Ltd., Shenyang, China
a
wuhl@mail.neu.edu.cn,
b
zhouruo.yi@qq.com,
c
flyingegg.ke@gmail.com
Keywords: Named entities recognition, Entity attribute, Semantic knowledge database.
Abstract. To construct the entity attribute semantic knowledge database in computer field, we need to
achieve the relationship between the entities and attributes. That requires to identify the
computer-named entities that present in the real text. Moreover, the verb collocation templates that
describe the relationships would be achieved. In this paper, the necessary knowledge to recognize
entities would be integrated into a generic framework by using entity-attribute concept. Thereby, the
rules of entity recognition would be simplified. We transform the named entities recognition process
of computer entities into an labeling process. For the given text to be processed, match the possible
brand words or serial words driven by the brand attribute value and the series attribute value. Then the
model sequence or the abstract entity suffix can be matched and marked in the text which successfully
marked the brand or series. Finally, match the results of the annotation with the recognition rules, and
output the marking sequence which accord with the rules as computer entity word. Proceed from the
idea of entity-attribute- framework, the name of an entity is the combination of the word mapping of
the entity's particular attribute value and the word mapping of the conceptual entity to which the entity
belongs. By writing the specified entity naming rules in such way, it is possible to organically
integrate the rules with the instantiation of supporting rules into the knowledge network centered on
entities, instead of forming irrelevant dictionary knowledge that is only isolated for specific tasks only.
Experimental result showed that the system achieved the F1 measure of 86.1%.
1. Introduction
To construct the entity attribute semantic knowledge database in computer field, we need to achieve
the relationship between the entities and attributes. That requires to identify the computer-named
entities that present in the real text. Moreover, the verb collocation templates that describe the
relationships would be achieved.
The traditional methods to recognize named-entities, whether rule-based or statistics-based,
organize and integrate the knowledge in different forms. But the knowledge isolates the entities. The
rule-based method meticulously divides the entity-word units. But the word units are the external
resources that are independent of the entities, just like dictionaries. The statistics-based methods only
statistically integrate the probability models from the high quality of the marked training resources.
The models deviate from the essence of language. In this paper, the necessary knowledge to recognize
entities would be integrated into a generic framework by using entity-attribute concept. Thereby, the
rules of entity recognition would be simplified.
2. The Process of Entity Recognition
The process of named entities recognition in computer field is shown in Fig.1. We transform the
named entities recognition process of computer entities into a labeling process. For the given text to