Artificial Intelligence

  • 1.  Unsupervised Clustering Approach for Twitter Sentimental Analysis: A Case Study for George Floyd Incident

    Posted Jun 17, 2020 12:07:00 PM

    Catastrophic national events have a ground effect on the lives of national citizens or even on people around the globe. The consequences of such a scenario are experienced in diverse and multiple ways. The outcome and reaction of public can be felt through media, news reports and social media. Performing sentiment analysis is vital which can be used to find out the public review about a product or ongoing events in the world. Public can easily and efficiently express their perspectives and ideas on a wide variety of topics like events, services, and brands via social networking websites. Social networks especially Twitter is continuously updated with public views, expressions, and opinions. In this I have performed twitter sentimental analysis to review public opinion about George Floyd incident using Twitter data. Text mining and sentimental analysis are used Text mining and sentiment analysis to analyze unstructured tweet text to extract positive and negative polarity about this incident. I found out that majority of the people have attitude towards this incident by using 3 hashtags and overall data.



    Tweet

    Polarity

    can you imagine feeling so empowered, that you allow yourself to be recorded while you commit murder in broad day light in front of several witnesses??? #blacklivesmatter

    Positive

    rest in peace # icantbreathe #georgefloyd

    Positive

    it is heartbreaking & terrifying living in a country where i would not call the police if i needed help, in fear that someone in my family could be wrongfully killed. #minneapolis #philandocastille #centralparkkaren #minneapolispolice #blacklivesmatter #icantbreathe

    Negative

    im so at a loss for words. this is just so hard to watch. #georgefloyd #rip! I hope all them cops a painful slow death fr

    Negative



    ------------------------------
    Balaji Karumanchi

    ------------------------------


  • 2.  RE: Unsupervised Clustering Approach for Twitter Sentimental Analysis: A Case Study for George Floyd Incident

    Posted Jun 19, 2020 07:40:00 AM
    FYI - For those who would like to know more about unsupervised natural language processing (NLP) and word vectorization for analyzing sentiment, this URL leads to a data science article written by Rafal Wojcik that helps explain the topic by way of method discussion and example.

     https://towardsdatascience.com/unsupervised-sentiment-analysis-a38bf1906483


    Depending on your level of interest, the overview and explanation is in the first few paragraphs of the article, and the remainder is a deeper discussion of applying unsupervised NLP methods.

    Mark Y.

    ------------------------------
    Mark Yanalitis
    ------------------------------



  • 3.  RE: Unsupervised Clustering Approach for Twitter Sentimental Analysis: A Case Study for George Floyd Incident

    Posted Jun 23, 2020 10:39:00 AM
    I think understanding sentiment in AI is going to be huge when it comes to data collection. At the back end of this too we still will have to address bias within the collection efforts.

    ------------------------------
    Sean Heide
    Research Analyst
    CSA
    ------------------------------



  • 4.  RE: Unsupervised Clustering Approach for Twitter Sentimental Analysis: A Case Study for George Floyd Incident

    Posted Jun 23, 2020 11:38:00 AM
      |   view attached
    Thank you Sean for the feedback. It took 24 hrs time for me to scrap the data from Twitter. The dataset is curated using scrapper built with Python. I retrieved 8,86,579 tweets by using 5 hashtags (blacklivesmatter, georgefloyd, icantbreathe, riots) over a span of 11 days from 25 May, 2020 to 4 Jun, 2020. The columns in datasets include username, tweet text, date of tweet and link to the tweet. The dataset is further divided into four sub-datasets which are created using top 2 hashtag (blacklivesmatter and George Floyd). Whereas to give insight about public reaction to protests, tweets related to protest are separated for tweet frequency analysis.

    All - I attached a complete research paper on "An Unsupervised Clustering Approach for Twitter Sentimental Analysis"..

    Feel free to provide your feedback. 


    ------------------------------
    Balaji Karumanchi
    Sr Manager
    Natsoft Corporation
    ------------------------------

    Attachment(s)



  • 5.  RE: Unsupervised Clustering Approach for Twitter Sentimental Analysis: A Case Study for George Floyd Incident

    Posted Jun 23, 2020 12:13:00 PM
    Thanks Balaji! This is super fascinating. Perhaps this could be used to compile trends in data security from social media as well? Would be an interesting project to scan for topics on a daily/ weekly basis.

    ------------------------------
    Sean Heide
    Research Analyst
    CSA
    ------------------------------



  • 6.  RE: Unsupervised Clustering Approach for Twitter Sentimental Analysis: A Case Study for George Floyd Incident

    Posted Jun 23, 2020 10:38:00 AM
    This is extremely interesting on the polarity scale of catching negative and positive outlooks through social media. Are there other instances you could share where a media source has used something like this to sway party's?

    ------------------------------
    Sean Heide
    Research Analyst
    CSA
    ------------------------------