Different Aggregation Strategies for Generically Contextualized Sentiment Lexicons

Stefan Gindl

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Sentiment detection has gained relevance in the last years
due to the vast amount of publicly available opinion in the form of Web
forums or blogs. Yet, it still suffers from the ambiguity of language,
lowering the efficacy and accuracy of sentiment detection systems. Thus,
it is important to also invoke context information to refine the initial
values of sentiment terms. Moreover, domain-independence is desirable
to avoid using a topic determination beforehand. This work investigates
strategies for extracting non-generic features to be integrated into a so called
contextualized sentiment lexicon, capable of getting the context
correctly and assigning sentiment terms the proper sentiment value. The
proposed approach will be applied in an online-media aggregation and
visualization portal, covering a vast number of news media sources.
Original languageEnglish
Title of host publicationDyNaK 2010 - Dynamic Networks and Knowledge Discovery
Subtitle of host publicationProceedings of the 1st Workshop on Dynamic Networks and Knowledge Discovery
EditorsRuggero Pensa, Francesca Cordero
Place of PublicationBarcelona, Spain
Pages89-100
Number of pages12
Publication statusPublished - 2010
EventDynamic Networks and Knowledge Discovery - Barcelona, Barcelona, Spain
Duration: 24 Sept 2010 → …

Conference

ConferenceDynamic Networks and Knowledge Discovery
Abbreviated titleDyNaK 2010
Country/TerritorySpain
CityBarcelona
Period24/09/2010 → …
Other1st Workshop on Dynamic Networks and Knowledge Discovery

Fingerprint

Dive into the research topics of 'Different Aggregation Strategies for Generically Contextualized Sentiment Lexicons'. Together they form a unique fingerprint.

Cite this