[Air-L] NLP for sentiment analysis of social media comment

[James Danowski]

My approach, SENET, goes beyond dictionaries to trace shortest paths from
4,500 sentiment words to the specified target word and computes positivity
and negativity values and ratios.

James Danowski
Communication and Technology Sciences

On Sat, Jul 14, 2018, 10:15 AM Astvansh, Vivek <astvansh at iu.edu> wrote:

> Hi, Nina:
>
> Yes; some scholars (particularly in the past, when field-specific
> dictionaries, such as one in finance, were not available) used their own
> supervised machine-learning method. This method comprised: (a) using human
> beings (e.g., MTurkers) to classify text into positive, negative, and
> neutral sentiment, (b) using this human classification to train a
> classifier (Lasso or support vector machine), and lastly, (c) using the
> trained model to classify the holdout sample. As you can guess, this is
> quite some work and perhaps not required when the relevant dictionary is
> readily available and you have a program that uses this dictionary to
> classify text.
>
> You can search Google Scholar to find articles that use human annotations
> and machine-learning classifiers and follow this supervised
> machine-learning approach.
>
> Best wishes!
> Vivek Astvansh
> Assistant Professor of Marketing
> Kelley School of Business, Indiana University Bloomington
>
> ________________________________________
> From: Air-L <air-l-bounces at listserv.aoir.org> on behalf of Nina Lasek <
> Nina.Lasek1 at hotmail.com>
> Sent: Saturday, July 14, 2018 10:51 AM
> To: air-l at listserv.aoir.org
> Subject: [Air-L] NLP for sentiment analysis of social media comment
>
> Dear all,
>
> most sentiment analyses of social media comments I know use
> dictionary-based approaches (e.g. sentigstrength). However, I am wondering
> if researchers also use Natural Language Processing approaches to examine
> the sentiment of social media comments; I am still an absolute beginner in
> this field, hence I would be very glad if somebody could point me to useful
> papers/tutorials/software for doing NLP based sentiment analyses?
>
> Many thanks,
> Nina
> _______________________________________________
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at:
> http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
>
> Join the Association of Internet Researchers:
> http://www.aoir.org/
> _______________________________________________
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at:
> http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
>
> Join the Association of Internet Researchers:
> http://www.aoir.org/
>