Social Media Analytics Platform
The requirement was to develop a social media analytics platform that provides deep insight into popularity or market value of entities across different domains in the social community. Data from different social media platforms, news websites, blogs etc should be analysed.
- Extracting the most relevant data or images from blogs and news website
- Information extraction from images, logos with accuracy
- Sentiment analysis on multilingual posts.
- After analysing the blogs and news websites a regular pattern was extracted and a custom web crawler was developed that makes use of these regular patterns to extract relevant data.
- TensorFlow was utilised for information extraction from images by image classification to provide 80% accuracy.
- Utilised word2vec to vectorise the posts and for multilingual support different language models were employed.
- Developed a social media analytics platform that shows the global, domain specific and sub domain based ranking of entities in sports, beverages or fashion domains. The entities, from their dashboard can identify their influencers which helps them in getting higher ranks. The affinity of a person towards a characteristic, theme, otherentities, hashtags can be analysed and visually represented in a bubble format.
- Business can know where to invest on,whom to sponsor for a sports event or find doppelgangers at affordable prices.