Intercomprehension in Retrieval
| Autor: | Hussein Suleman, Catherine Chavula |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
050101 languages & linguistics
Computer science business.industry 05 social sciences Intelligibility (communication) computer.software_genre Preference Ranking (information retrieval) Resource (project management) 0501 psychology and cognitive sciences Relevance (information retrieval) Artificial intelligence 0509 other social sciences 050904 information & library sciences business Intercomprehension computer Natural language processing |
| Zdroj: | CHIIR |
| DOI: | 10.1145/3343413.3377954 |
| Popis: | Most of the Web is published in languages not accessible to many potential users who are only able to read and understand their local languages. Prior research has focused on using translation to provide users with information written in other languages, yet there are still many languages with little or no such resources. In this paper, we propose the use of intercomprehension -- a form of communication in which speakers of two different languages communicate using their own languages, mainly due to similarities between the languages. Accordingly, we conducted a user study on six Sub-Saharan languages to explore: user interaction behaviour in a retrieval environment where intercomprehension is expected; to investigate the usefulness of search results that assume intelligibility and relevance; and investigate affective episodes associated with intercomprehension in retrieval through retrospection. Although intercomprehension may come with a cost to understand unfamiliar languages, our results show that user preference of ranking of search results in related languages incorporates intelligibility. Our findings also suggest that intercomprehension is useful in retrieval for related languages -- users are able to identify relevant documents as well as complete search tasks by applying intercomprehension. However, the negative emotions or frustration associated with intercomprehension suggest that this type of interaction should be used in extreme cases where there are no or very few relevant documents in the language of the query. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|