Please enable JavaScript.
Coggle requires JavaScript to display documents.
Building Sentiment Lexicons for All Major Languages (Information (Graph…
Building Sentiment Lexicons for All Major Languages
Information
Methodology
Building high-quality sentiment lexicon for 136 major languages
Polyglot project
Source
Wiktionary
Machine Translation
Transliteration Links
WordNet
Graph Propagation
Through semantic links in our knowledge graph, words are able to extend their sentiment polarities to adjacent neighbors.
We experimented with both graph propagation algorithm (Velikovich et al., 2010) and label propagation algorithm (Zhu and Ghahramani, 2002; Rao and Ravichandran, 2009)
What is
Graph propagation
Meta
Goals
Problem
Sentiment analysis in a multilingual world remains a challenging problem, because developing language-specific sentiment lexicons is an extremely resource intensive process
Such lexicons remain a scarce resource for most languages.
Objective
Result
Contribution
New Sentiment Analysis Resources – We have generated sentiment lexicons for 136 major languages via graph propagation which are now publicly available
Large-Scale Language Knowledge Graph Analysis – We have created a massive comprehensive knowledge graph of 7 million vocabulary words from 136 languages with over 131 million semantic inter-language links,
Extrinsic Evaluation – We elucidate the sentiment consistency of entities reported in different language editions of Wikipedia using our propagated lexicons