Chinese Sentiments on the Clouds: A Preliminary Experiment on Corpus Processing and Exploration on Cloud Service

Copyright 2012 by Shu-Kai Hsieh, Yu-Yun Chang, and Meng-Xian Shih 26th Pacific Asia Conference on Language,Information and Computation pages 491–497 Chinese Sentiments on the Clouds: A Preliminary Experiment on Corpus Processing and Exploration on Cloud Service Shu-Kai Hsieh, Yu-Yun Chang, Meng-Xian Shih Graduate Institute of Linguistics, National Taiwan University No. 1, Sec. 4, Roosevelt Road, Taipei, 10617 Taiwan shukaihsieh@ntu.edu.tw, {june06029;simon.xian}@gmail.com Abstract This study aims to propose a novel pipeline architecture in building and analyzing large- scaled linguistic data on the cloud-based envi- ronment, an experimental survey on Chinese Polarity Lexicon will be taken as an exam- ple. In this experiment, data are evaluated and tagged by applying crowd sourcing ap- proach using online Google Form. All the data processing and analyzing procedures are com- pleted on-the-fly with free cloud services au- tomatically and dynamically.The paper shows the advantages of using cloud-based environ- ment in collecting and processing linguistic data which can be easily scaled up and effi- ciently computed. In addition, the proposed pipeline architecture also brings out the poten- tials of merging with mashups from the web for representing and exploring corpus data of various types. 1 Introduction With the emergence of huge amount of web data available in recent years, corpus linguistics as well as other related empirical fields such as the collect- ing and processing of language resources, and their evaluation are facing with the greatest challenges ever. The spread of corpus and lexical resources in linguistics has been led to a great level of theo- retical survey and enhanced the empirical founda- tion, not only with respect to sampling and annota- tion, but also with exploratory data analysis. How- ever, more recently there have been long discussions about what the current state of art in corpus linguis- tics fails to do, which can be pinpointed at least in two respects: (1) the lack of socio-cultural (meta-) information reflected in the data, is incompetent for pragmatic usages and discourse analysis; (2) rather skewed with data in the public domain, heterogene- ity of (individualized) language usages and develop- ment is not able to be traced. With the advanced technological progress in data availability with storage and computing ability, the issues mentioned can be tackled to a great extent. We take it as the turning point for corpus-based lin- guistics to transform into a data-intensive and cloud- based linguistics. In light of that, we want to ex- plore the transformation viability in this paper. As a first step, we present a novel pipeline architecture to build Chinese Polarity Lexicon on the cloud en- vironment by taking the data from the web as re- source. Polarity lexicon contains sentiment-bearing words and phrases, encoded with polarities to each word or phrase, usually either assigned as positive or negative. The study of polarity lexicon has attracted much attention in recent years for classifiers to train on the lexical dataset, and is becoming important for applications such as Sentiment Analysis and Opin- ion Mining. For the purpose of constructing automatic iden- tifying and …

Content truncated to 500 words