Sentiment Analysis

[vishu7560]

Hi

I am new to Knime.

I have one senerio where i need to extract data (all the comments in last 1 months for politician like Narendra modi and Rahul Gandhi)from the below site.

http://economictimes.indiatimes.com/

i want to do sentiment Analysis on that data(Comments).

Please anyone help me on this.

Thanks,

Vishvajit

[kilian.thiel]

Hi Vishvajit,

to crawl a webpage like a forum the Palladian KNIME extension is very useful (see: http://tech.knime.org/community/palladian). The extension provides nodes to download/crawl the content of a webpage. In this white paper we used it to crawl the KNIME forum and analyze it afterwards: http://www.knime.org/files/knime_web_knowledge_extraction.pdf.

For sentiment analysis you can use the Textprocessing extension. See: http://tech.knime.org/knime-text-processing for examples and documentation. Here: http://www.knime.org/files/knime_text_processing_introduction_technical_report_120515.pdf you find an introduction to the extension. In this white paper: http://www.knime.org/files/knime_social_media_white_paper.pdf we used the extension to analyze the slashdot form, including sentiment analysis.

To do sentiment analysis in KNIME basically you need a positive and negative word list. Use the Dictionary or Wildcard Tagger on the documents with your pos and neg dictionaries, to assign sentimentr tags to the words. Then you can count the frequencies of pos and neg words and create a sentiment rating for a entry, e.g. forum post or reply.

Hope this helps.

Cheers, Kilian