ETRI-Knowledge Sharing Plaform

ENGLISH

성과물

논문 검색
구분 SCI
연도 ~ 키워드

상세정보

학술대회 The Development and Verification of a Similar Crimes Search System in Korean
Cited 0 time in scopus Download 1 time Share share facebook twitter linkedin kakaostory
저자
박원주, 장광호, 백명선, 이용태
발행일
202110
출처
International Conference on Information and Communication Technology Convergence (ICTC) 2021, pp.1233-1235
DOI
https://dx.doi.org/10.1109/ICTC52510.2021.9621149
협약과제
21HR4200, 위험 상황 초기 인지를 위한 ICT 기반의 범죄 위험도 예측 및 대응 기술 개발, 이용태
초록
In this paper, NLP based technique to search past crimes according to crime similarity is introduced. In the case of crimes, criminal method similarity is more important than text similarity. Therefore, we intend to analyze the criminal method similarity by extracting the words from document describing the crime, recognizing the named entity and tagging relational information of the words. Especially, by setting weights on each information, we search for similar crimes that fit a user's intentions. When a new case is received, similar past cases are searched for by analyzing the similarity with words, relational information and their weights. In particular, the search results are evaluated based on NDCG to measure the performance.
KSP 제안 키워드
Search results, named entity, search system, text similarity