Tunisian dialect Wordnet creation and enrichment using web resources and other Wordnets

Abstract

In this paper, we propose TunDiaWN (Tunisian dialect Wordnet) a lexical resource for the dialect language spoken in Tunisia. Our TunDiaWN construction approach is founded, in one hand, on a corpus based method to analyze and extract Tunisian dialect words. A clustering technique is adapted and applied to mine the possible relations existing between the Tunisian dialect extracted words and to group them into meaningful groups. All these suggestions are then evaluated and validated by the experts to perform the resource enrichment task. We reuse other Wordnet versions, mainly for English and Arabic language to propose a new database structure enriched by innovative features and entities.

Contributions to the automatic processing of the user-generated Tunisian dialect on the social web

“…Work on the construction of TD ontologies includes the Wordnet 'TunDiaWN' proposed by Bouchlaghem et al (2014). The Wordnet was constructed from a corpus named multi-source Tunisian dialect corpus (MultiTD), collected from various sources (social networks, written plays, dictionaries, transcription of speeches, etc.).…”

“…The followed approach results in a unique form of each transcribed word and therefore cannot cover the written form of TD, especially the one that is daily produced on the social web, which is, as shown in Section 2, rich, varied and does not conform to specific rules or standards. Other TD resources were built using existing MSA resources (Boujelbane et al, 2013b;Bouchlaghem et al, 2014). Although benefiting from MSA corpora and tools helps importantly overcoming the lack of LRs, this approach does not consider TD language productions that are not MSA derived, especially when dealing with the TD used on the social web.…”

“…This is why many researches in the natural language processing (NLP) field have been carried out in recent years, in order to deal with the ADs, especially when they are written on the social web. Some of them focused on constructing corresponding language resources (LRs) (Diab et al, 2000;Zaidan and Callison-Burch, 2011;Meftouh et al, 2012;Boujelbane et al, 2013a;Bouchlaghem et al, 2014;Younes and Souissi, 2014), others tackled their automatic processing (Al-Sabbagh and Girjuh, 2012;Chalabi and Gergers, 2012;Darwish, 2014;Masmoudi et al, 2015;Younes et al, 2016;Aridhi et al, 2017).…”

A survey on author profiling, deception, and irony detection for the Arabic language

“…Cavalli‐Sforza et al () created an Iraqi Arabic WordNet using an English‐Iraqi dictionary and the modern standard Arabic version of WordNet as well as the English WordNet. Moreover, a Tunisian dialect WordNet was built in Bouchlaghem, Elkhlifi, and Faiz () starting from a Tunisian corpus.…”

Natural Language Processing for Dialectical Arabic: A Survey

“…In (Cavalli-Sforza et al, 2013) an Iraqi Word-Net is presented based on the MSA WordNet, the English WordNet, and an English-Iraqi dictionary. A Tunisian dialect WordNet was built in (Bouchlaghem & Elkhlifi, 2014) starting from a Tunisian corpus.…”