110 likes | 121 Views
Linked Open Data Query Based on Natural Language. Jie Liu, Wei Li, Liming Luo, Jianshe Zhou, Xu Han, Jinsheng Shi Capital Normal University. Wentao Ding. Chinese Journal of Electronics Vol.26 , No.2, Mar. 2017. Conversion system based on Entity-property-modifier (EPM ).
E N D
Linked Open Data Query Based on Natural Language Jie Liu, Wei Li, Liming Luo, JiansheZhou, Xu Han, Jinsheng Shi Capital Normal University Wentao Ding Chinese Journal of Electronics Vol.26, No.2, Mar. 2017
Conversion system based on Entity-property-modifier (EPM) • Unsupervised query parsing method • Classify the NL words into five types • Rule based triple extraction • Map the triple patterns • Mapping by predicate & entity index lookups • Select vocabulary based on semantic & syntax similarity
Annotation • POS based words classification • e.g.
Query Analysis • Query Type and focus
Query Analysis • Triple construction • Dependent Relationship -> DAG • Only semantic dependencies associated with key words
Query Analysis • Triple instantiation
Query Analysis • Give me all actors who were born in Berlin. • Result
Resource Mapping • For entity, • Use Wikipedia to locate the resource in homepage • A special index is provided for mapping country. • For modifier constraint • Triple instantiation rules • For concept and relation/property • Use the WordNet to get its synonyms, hypernyms, and relative synset, and their corresponding indexes are used to map the resources
Resource Mapping • Select candidates • A semantic score to get the top-K prior candidates Lin’s semantic similarity method on WordNet a fixed value in “our common index”
Experimental Result • Standard test set provided by QALD-4
Q & A • Thanks for listening.