Do you know what exactly did the Union Finance Minister Dr Nirmala Sitaraman stressed on the recent budget speech, the longest speech in the budget history of India? Well, it was tax and agriculture. 

A research team at the Indian Institute of Technology Madras (IITM) has analysed the Union Budget using a recently developed Artificial Intelligence-driven tool. The tool generated a theme-based word cloud that comprises prominent topics mentioned by the finance minister. Size and position of the topic in the word cloud are proportional to the emphasis given for the same in the budget. The analysis revealed that the Union Budget stressed more on tax and agriculture, with the word “farmers” taking centre stage, followed by “welfare schemes” and “new schemes.” 

While tax and agriculture got 19.4 per cent and 15. 3 per cent attention in the budget speech; economic schemes, welfare schemes, technology, healthcare and growth also got notable attention. Issues related to farmers got maximum attention (45.5%) followed by GST (28.6%). Direct Tax, Tax concession, Live stocks, etc. were particularly stressed in the speech. 

The RISE Group (Reconfigurable and Intelligent Systems Engineering group) in the Department of Computer Science and Engineering, IIT Madras, are behind this analysis of the Union Budget speech. The researchers employed an AI-based automated analysis employing NLP (Natural Language Processing) techniques on the budget speech. The individual themes were analysed again to find out that GST and farmers got more mentions in the speech. 

Open source based AI tool 

The open source based analytics stack, DataLibre and open source natural language processing techniques were used to develop the AI speech analytics tool. The tool was developed by a team of researchers including Prof. V. Kamakoti, R. Kaushik and P. Nagarajan. The team started developing the tool three years back using open-source software. 

“To analyse any text or document, you have to first provide the English text to the system. The software then analyses it and identifies the themes in it. The output will be in the form of a word cloud. Each theme can then go for fine-grain analysis, and the system will find out the concentration for each theme. Based on that we can come up with inferences, “says Kamakoti. 

The tool also helps in doing sentiment analysis. 

Industry application 

According to the researchers, companies and businesses can use this tool to collect customer surveys and to analyse customer feedback. “Suppose you have a large number of customer feedback and you want to understand their sentiments, it is not easy to go through each and everyone to analyse them. This AI tool can analyse them and come up with the major sentiments,” says Kamakoti. The AI tool could also be used to analyse annual reports of companies, hospitals and businesses, and can produce meaningful information in a nutshell. 

The DataLibre team has done a similar NLP-Sentiment analysis on the first fifty episodes of Prime Minister Modi’s “Mann Ki Baat” programme before. Currently, the tool works only with texts in English. The team is also working on enhancing the tool to further quantify the sentiments, for instance, the number of positive statements, negative statements, statements of concerns, and so on.