6533b82dfe1ef96bd1291f65
RESEARCH PRODUCT
An Approach to Enhance Chatbot Semantic Power and Maintainability: Experiences within the FRASI Project
Giovanni PilatoAlberto MachìAgnese AugelloSalvatore Gagliosubject
Settore ING-INF/05 - Sistemi Di Elaborazione Delle Informazionieducation.field_of_studyComputer sciencebusiness.industryPopulationComputingMilieux_PERSONALCOMPUTINGOntology (information science)computer.software_genreSemanticsDomain ontologies; In-buildings; Industrial projects; Inference process; Preprocess; Question answering systems; Question-answer pairsChatbotSemantic networkDomain (software engineering)Knowledge-based systemsArtificial IntelligenceArtificial intelligenceUser interfacebusinesseducationcomputerNatural language processingdescription
The paper illustrates the implementation and semantic enhancement of a domain-oriented Question-Answering system based on a pattern-matching chat bot technology, developed within an industrial project, named FRASI. The main difficulty in building a KB for a chat bot is to handwrite all possible question-answer pairs that constitute the KB. The proposed approach simplifies the chat bot realization thanks to two solutions. The first one uses an ontology, which is exploited in a twofold manner: to construct dynamic answers as a result of an inference process about the domain, and to automatically populate, off-line, the chat bot KB with sentences that can be derived from the ontology, describing properties and relations between concepts involved in the dialogue. The second one is to preprocess user sentences, and to reduce them to a simpler structure that can be referred to existing elements of the chat bot KB. The enhanced symbolic reduction of user questions and the automatic population of question templates in the chat bot KB from domain ontology have been implemented as two computational services (external modules). © 2012 IEEE.
year | journal | country | edition | language |
---|---|---|---|---|
2012-09-01 | 2012 IEEE Sixth International Conference on Semantic Computing |