Innodatatics rated among the "20 Most Promising Data Analytics Solution Providers - By CIO Review"
Social Media Analytics

Social Media Analytics

Web-based Media Analytics will help draw experiences from unstructured information, like Text Mining and Natural Language Processing strategies. To play out the positive and negative things, i.e., Sentimental Analysis, where the globe is communicating about such countless things. Nimble deviation from normal will help counter the abrupt occurring of occasions, which will cost the extreme damage. The organizations can get beneficial simply by utilizing more intelligent by forestalling highlights. Every one of the items and interests will be suggested utilizing clients' web-based media use, and the rundown will go perpetually.

Case Study 1: Want to know on how Sentiment Analysis is performed from Twitter’s unstructured data?
Business Problem Statement

With the rise of digitalisation, the usage of social media access reach of the average citizen has multiplied. Advent of technology not only comes with the advantages, but also the disadvantages. Many people who have access to the internet do not restrain from giving to-the-point feedbacks, and are not at all shying away. But sometimes, these reviews/feedbacks are given only because of the unhealthy competition. At times, this is creating a lot of trouble to the genuine products/manufacturers, risking them to drop the plans of manufacturing those products. It also results in dropping of rating of those products.

Business Challenges

  • Increase in the complexity in deriving insights out of Opinionated Text
  • Very hard to find the genuinity of the opinion on social media such as Twitter
  • Complexity in reading unstructured data
  • Relying on external surveys, consultants and other related groups
  • Heavy cost is incurred in finding the sentiment associated with product

Business Solutions

  • Text Mining/Web mining performed on the twitter handle pertaining to the product to get accurate image of the brand/product
  • Genuinity of the reviews are derived using the most advanced techniques involved in Natural Language Processing (NLP)
  • Sentiment Analysis is performed at the highest level of accuracy to infer the perception associated with the Brand/Product
  • Built classification model to increase the signal-to-noise ratio by classifying the tweets/comments on social media as very risky, risky, neutral, etc.
  • Drawing meaningful business insights based on word clouds, semantic networks, cluster word clouds & dendrograms

Business Impact

  • Innovative features are incorporated into the product
  • Scope for automation increased
  • Improvisation in Brand Image due to proactive redressal of issues
  • Around 10% improvement in Customer rating
  • Improved customer retention rate

Case Study 2: How risk sensing is performed on companies by extracting data from social media sites
Business Problem Statement

Living in the internet era & social media brings with it a flurry of new challenges. With the increase in rapid sharing of information (good, bad, ugly) about the companies & its products/services, the focus has seen a sudden surge. Critical comments are spreading at a breakneck pace, effectively "making mountains out of molehills." These messages blowing out-of-proportion is leading to a lot of fiasco & shaming the brand image to such an extent that reversal of the negative mindset of customers is taking years together in-spite of everything being ethically correct.

Business Challenges

  • Negative vibes on social media were rising, thereby impacting the brand image of the organizations in spite of decades of strong presence
  • Existence of too many online free–to–post forums where people can express, or debate about products, services, organizations, celebrities, etc.
  • Increase in risk of brand vulnerability because of increase in social media penetration
  • Insignificant issues being blown out of proportion because of fast outreach of social media

Business Solutions

  • Live streaming of social media data based on Twitter hash tags & Twitter handles and performing sentiment analysis to identify negative words
  • Capturing data from top 2 social media websites to identify anomalies by pattern recognition
  • Built classification model to increase the signal-to-noise ratio by classifying the tweets/comments on social media as very risky, risky, neutral, etc.
  • Drawing meaningful business insights based on word clouds, semantic networks, cluster word clouds & dendrograms

Business Impact

  • Increased customer satisfaction by 6%
  • Reduction in number of times the product, service, organization, person was in media for negative reasons
  • Improved brand image due to proactive redressal of issues
  • Increased in customer retention rate
  • Increased bottom-line & top-line
  • Identification of issues beforehand lead to innovation

Case Study 3:How to increase the probability of ‘click-through rate’ of ads posted on social media
Business Problem Statement

World is taking huge steps in terms of automation. The amount of time invested on screening through the profiles is very huge and is becoming a tedious task for the resources involved in this activity. This leads to cost to the company and dissatisfaction of the resource, inturn leading to quality issues.

Business Challenges

    • Customers most often browse the websites, which appear on the first page of the search engine, leaving heavy competition for usage of relevant keywords, thereby triggering paid advertisements
    • Decrease in the click-through rate making the ad irrelevant & impacting the quality score of ad, thereby not showing the ad at all
    • Decrease in the transaction conversion rate leading to increased cost per click but customer not buying the product
    • Lack of statistical way of identifying the keyword, which will increase the chances of lead conversion

Business Solutions

  • Developed statistical techniques to identify the keywords, which will increase the ad quality & thereby chances of clicking on the ad
  • Statistically analyzing the ad quality & landing page quality & determining on which of these has to be strategically improved
  • Calculating probability of conversion of leads who click on the ad by multiplying click-through rate with transaction conversion rate
  • Building a prediction model to determine what parameters are relevant for increasing the conversion rate of the leads

Business Impact

  • Increased ranking of the advertisement on search engine
  • Improvement in revenue due to lead conversions
  • Increased profit margin after removing the cost per click expenses
  • Increase in number of website visitors, which contributed to better branding of the organization

Case Study 4: How to combine ‘Arcs & Emotion Mining’ for anomaly detection
Business Problem Statement

Too many movies are being directed without knowing about what really catches the viewers spellbound. With time, the budget spent on movie production is inflating beyond imagination & a flop at the box office would mean that the entire crew will be pushed out of race for atleast a few years. There are many scenarios where the entire production houses were shut down because of producing flop movies. Careers of the cast is at stake if they feature in a flop movie. Lack of unique, new, interesting storyline is always adding to the misery of the industry.

Business Challenges

  • Lack of different story lines for movie production due to vast number of movies being produced
  • Only a small fraction of new movies break even, leading to huge losses
  • Lack of insights on movie tastes due to ever changing demographics of the new population
  • Only a small segment of film professionals are able to come up with blockbusters, on which the whole industry piggyback rides

Business Solutions

  • Used Arcs & Motions of natural language processing to understand the various emotions of the movies across the time series duration of the movie
  • Analyzing the positive & negative emotional valence of all the blockbuster movies to understand about, which emotion at specific time periods will constitute to highest viewership
  • Combining the social media reviews to understand on what aspects the movies are being liked, leading to a much superior story-line for sequel/prequel than the present version

Business Impact

  • Reduction of the routine boring story-lines thereby reducing losses
  • Increasing the probability of success of movies by directly accommodating the opinions of the viewers
  • Increasing the number of newer ideas for the film industry crew
  • Increasing the opportunity for new crew to venture into film industry

Case Study 5: Analytics on Political party representatives
Business Problem Statement

Citizens are reporting to posting messages on social media & web to vent out the frustration or happiness associated with the daily activities going around. There is no transparency on how many promises were done by political party members at the time of election. Lack on clarity on the performance of the elected representatives leading to some sections misguiding the society with false claims.

Business Challenges

  • Dissatisfied & frustrated citizens resorting to social media to crib about the unhappiness
  • Top leadership of Political parties are unable to identify the rogue elected representatives
  • Unable to identify the good & bad of the different locations under the jurisdiction of elected representatives
  • Unable to predict on whom should be given the party ticket based on popularity

Business Solutions

  • Scarped the web & social media to extract the data pertaining to the various political parties, elected representatives & performed Natural Language Processing to determine the issues which people are happy/unhappy about
  • Used Support Vector Machine on the cleansed data to determine the anomaly of the positive/negative trend
  • Predicting the popularity score to determine the success of the person winning elections

Business Impact

  • Quick review of performance of political parties
  • Quick review of politicians’ performance to citizens & party leadership
  • Great insight for news channels for exit poll analysis
  • Increase citizen involvement & healthier society because of transparency

Contact Us