[Corpora-List] SentiLecto: New approach for Entity-based Sentiment Analysis in Spanish (work in progress)

[Fernando Balbachan]

Spanish is a particularly challenging language for syntax analysis, for it is a free-order constituency language. Full parsing in Spanish is very tough, as it is not obvious where to find high level syntax functions as subject, direct object, etc. Moreover, Spanish deals with a morphologically rich system in pronouns, agreements, etc. which makes the task harder.

SentiLecto http://tecnolecto.com/sentilecto is our Spanish Sentiment Analysis solution at Tecnolecto. This Entity-based Sentiment Analysis (ESA) solution yields a granular representation for the sentiment values involved in each opinion. Unlike other approaches, this solution can deal with polarity shifting in the same sentence ('I like chocolate but I hate strawberry ice-cream') or even within embedded clauses ('Norwegians, who are an aggresive People, export the exquisite herring'). SentiLecto better represents the assumptions whereby the entities involved in the opinion are syntactically mapped onto SVO (subject-verb-object) slots for their sentiment assignments: 'Mary hates John' (2 entities but only the object has a negative presentation) vs. 'Mary defames John' (the same 2 entities but only the subject has negative presentation).

SentiLecto leans on outstanding linguistic features such as: passive/active voice transformation, anaphora resolution and co-reference chains, modality treatment and accurate verb scripts for all verbs in Spanish, even with different pronominal cases (for example, 'destacar' 'to appraise something' vs. 'destacarSE' 'to highlight oneself from the rest')

Try it out and behold the results! http://tecnolecto.com/sentilecto 

Fernando Balbachan
Senior Computational Linguist
Tecnolecto Group
http://tecnolecto.com 

Disclaimer: So far, we have released the syntax analysis, including passive/active voice transformation, clause extraction, and anaphora resolution but we are working on modality and sentiment analysis. We will adapt this approach to English in a near future.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/corpora/attachments/20140517/0adb2acf/attachment.htm>
-------------- next part --------------
_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora