Conversation Frames: Yet Another Contextual Dimension for IPAs
Autor: | Omar Saad Almousa, Hazem Migdady |
---|---|
Jazyk: | angličtina |
Rok vydání: | 2019 |
Předmět: | |
Zdroj: | Proceedings of the XXth Conference of Open Innovations Association FRUCT, Vol 622, Iss 25, Pp 25-31 (2019) |
Druh dokumentu: | article |
ISSN: | 2305-7254 2343-0737 |
Popis: | At first, we aimed at comparing the only two Intel- ligent Personal Assistants (IPAs) that support Arabic language, namely Siri of Apple and Salma of Mawdoo3. We compared them according to two features concerning the given answers: 1) Correctness and 2) Naturality. During our evaluation, we found out that both IPAs have a shortcoming that is present for other languages as well. This shortcoming is their inability to take into consideration the contextual information derived from consequent questions. To address this shortcoming, we propose a model that improves the naturality of an IPA without downgrading the correctness of its answers. |
Databáze: | Directory of Open Access Journals |
Externí odkaz: |