义原驱动下领域本体模型动态更新
Dynamic Updating of Domain Ontology Model Driven by Semantic Primitives
针对部分静态本体不能满足领域知识的动态更新需求与系统化表示的问题,提出一种义原驱动的本体模型构建与更新方法。首先,运用通用信息抽取(UIE)技术与依存句法分析技术,分别获取事件集合与成分三元组集合;其次,采用词性标注与名称相似性计算方法,对两组集合中特定元素判断词性与名称相似性,得到触发词匹配集合、事件论元匹配集合与未匹配集合;再次,义原相似度被计算以提取集合中候选词与辅助词集的公共义原,通过度中心性计算对义原排序,选择最合适义原作为预选的新增本体;最后,借助Protégé本体构建工具,以城市信息和地理兴趣点两个不同本体模型为例进行动态更新,验证了方法的有效性、鲁棒性与通用性。
Aiming at the problem that some static ontologies cannot meet the dynamic update requirements and systematic representation of domain knowledge, a semantic primitives-driven method for ontology model construction and update is proposed. Firstly, by using universal information extraction (UIE) technology and dependency syntactic analysis technology, the event relation set and the component triplet set are obtained respectively. Secondly, by using the methods of part-of-speech tagging and name similarity calculation, the part-of-speech and name similarity of specific elements in the two sets is judged to obtain the matching set of trigger words, the matching set of event arguments and the unmatched set. Then, the similarity of semantic primitives is calculated to extract the common semantic primitives of candidate words and auxiliary word sets within the set. These primitives are then sorted through degree centrality calculation, and the most suitable ones are selected as the preselected new ontology. Finally, with the aid of the Protégé ontology construction tool, dynamic updates are carried out by taking two different ontology models of urban information and geographical points of interest as examples, verifying the effectiveness, robustness and universality of the method.
domain ontology / ontology updating / semantic primitives / knowledge extraction / similarity
1. E1 = Ø, E2 = Ø, E4= Ø //初始化集合
2. FOR EACH ep IN EP DO
3. sim_trig = CALCULATE_SIMIL (sub, trig)
4. //计算主语与事件论元的相似度
5. sim_argu = CALCULATE SIMIL (sub, argu)
6. IF sim_trig >= THRESHOLD THEN
7. E1 = E1∪{eu} //添加到匹配触发词集合
8. ELIF sim_argu >= THRESHOLD THEN
9. E2 = E2∪{eu} //添加到匹配事件论元集合
10. //计算谓语与E1事件论元相似度
11. FOR EACH ea IN et.EVENT_ARGUMENTS DO
12. sim_obj_argu= CALCULATE_SIMIL (obj, ea)
13. IF sim_obj_argu > THRESHOLD THEN
14. E4=E1 \ {et}
15. BREAK
16. RETURN E1, E2, E4
| [1] |
|
| [2] |
白宇,田雨,王之光, |
| [3] |
张婉晨,郭黎,张毅, |
| [4] |
李墈婧,胡建军,吴迪, |
| [5] |
王恒,杨淑群.基于形式概念分析的制造业动态本体知识库构建[J].制造业自动化,2021,43(4):136-140. |
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
张茜.基于语义网的非遗元数据本体构建与可视化研究[J].江苏科技信息,2024,41(17):75-80. |
| [10] |
李慧敏,刘新贵,闫旭强, |
| [11] |
薛孟武,吴晓芳,黄振铭.面向电子对抗领域装备知识图谱的本体构建[J].空天预警研究学报,2024,38(5):331-335;341. |
| [12] |
|
| [13] |
刘群, 李素建.基于《知网》的词汇语义相似度计算[J].中文计算语言学,2002,7(2):59-76. |
| [14] |
程玉胜,梁辉,王一宾, |
| [15] |
崔卓,李红莲,张乐, |
| [16] |
黄振铭,吴晓芳,薛孟武.雷达知识图谱构建方法及应用[J].空天预警研究学报,2024,38(3):178-183. |
| [17] |
陆青梅.基于语义分析的网络舆情研究[D].武汉:武汉大学,2019:1-125. |
| [18] |
孙润志,于放.基于《知网》的词语相似度计算方法[J].计算机系统应用,2015,24(7):155-158. |
| [19] |
刘阳,付云洁,洪志佳, |
智慧地球重点实验室开放基金(KF2023ZD02-04)
/
| 〈 |
|
〉 |