Data Driven Decision Making

Procensol > Procensol Labs > Data Driven Decision Making

Brief summary

Are you ever stuck in reaction-mode when your mind tries to process the plethora of news, information and commentary presented every day, every hour, every minute on social media? Have you ever felt like you’re playing whack-a-mole when starting a conversation with a colleague or pitching an idea on a trending topic? Using artificial intelligence and machine learning, Procensol’s Acacio Barrado took just a few days to develop a Twitter Sentiment Analysis Bot with laser sharp accuracy to understand the emotional tone of a topic on Twitter. By utilising community editions of some highly trusted tools – UiPath, Twitter APIs and IBM Watson, Acacio has demonstrated how easy it can be to use bots and artificial intelligence to create a really practical application for a business, in this case, something that would be of great value to marketing, brand monitoring or customer experience teams.

  • test
Share:
Print
+100%-

The only limit to what we can create and innovate from this bot is our imagination.

Source: Acacio Barrado

Author Image

Acacio Barrado

Appian Consultant

About the Author:

Author Image

The idea for creating this tool was simple – imagine a situation where you are presenting on a topic, or launching a brand idea and you want to have a brief sense of the public perception about the topic in question based on the comments that they post on social media (Twitter in this case). Are the comments positive? Negative? Neutral?

The conventional way to quickly prep for a conversation and gain more confidence is to do a google search, talk to people and check out social media…in effect gathering up more and more information to inform your own brief and try and gain a sense of the public perception of the topic. In particular for social media, you try and manually process in your mind the likes, reactions, tweets and comments that social media users post on different platforms. There must be an easier way to analyse the sentiment, particularly of natural language!

Natural language is used daily to express an individual’s joy, happiness, anger or frustration on social media. Emojis are commonly used as an accentuated punctuation for most millennials. Sentiment analysis could perhaps be every marketing companies’ and political figures’ best kept secret. It is an automated process of understanding the emotional tone of a written opinion allowing parties with that information to address a response.

In this instance though I was not in need of anything fancy. I was purely on the hunt for a simple and accurate social media sentiment analysis tool for my personal daily use. (And I didn’t have budget to outlay for the off the shelf solutions already available)

So instead, I used community editions of UiPath, (a tool that is able to orchestrate the sentiment analysis process), Twitter APIs as the source of data and IBM Watson to analyse

the comments and prove the analysis. Voila – my Twitter Sentiment Analysis Intelligent Bot was built.

How to Use the Twitter SA Intelligent Bot:

When the bot is started, a pop-up display window appears for the user to insert a term or a phrase to be searched; the search accuracy is higher if the term uses a hashtag or a single word. However, short sentences also produce results. It is required of the user to insert an email to receive the sentiment analysis results.

With the information provided by the user, the bot starts to trigger the Twitter API’s to complete the authorisation process and retrieve the last 100 tweets that contain the search term. The bot stores the Twitter responses, filtering only the information that is required for the process, such as users, location and comments. Data is then stored in the bot’s internal tables and soon after, presented in an excel file.

Once the excel file is created, the bot gets the comments and starts to use the IBM Watson API’s. This starts with the authorisation process followed by analysing the text sentiment.

For each comment, the bot collects the response from IBM Watson and adds it to the Excel file. The normal response from Watson is one of the three: Positive, Negative or Neutral.

With the Excel file created, the Twitter Sentiment Analysis bot refreshes the pivot table in Excel in order to update the charts previously created.

Finally, the bot sends an email back to the user, using the email provided in the beginning of the process, attaching to this email an Excel file that contains all the data extracted from Twitter along with a chart showing the Comments and Sentiment Analysis.

Using my process automation knowledge and readily available robotics process automation and artificial intelligence systems I was able to create a simple tool that can provide useful data based on Twitter comments – and I did all this using free community versions – Imagine the possibilities with the full systems!

The only limit to what we can create and innovate from this bot is our imagination.

Author image

Acacio Barrado

Appian Consultant

About the Author:

Author Image