The central objective of this project is to reach out to millions of farmers of India using social media and machine learning.
Agriculture extension is vital to disseminate information on new agriculture techniques and agronometric practicies among the farmers. Even if we develop the best varieties of seeds but if we are not able to disseminate its information amongst the farmers, it can hardly have any impact. Once the farmers come to know about the advanced seed, only then will they demand it and use it. Also just using the seed may not be sufficient. It may need a different way of applying fertilizers or different watering time. The farmer needs to know about all this, and hence the value of extension.
Given the agriculture sector scenario in India, extension becomes a very challenging task. There are millions of small and marginal farmers and landless labor - their awareness levels are low, many live in remote areas and so on.
Traditionally our extension workers and field officers visit the fields and interact with farmers invillages and provide them training & advisory on best practices in farming and aspects of agriculture. Additionally, various training programs are organized by various agencies under the various schemes run by the agriculture department the details of which can be found at the department’s website.
In recent years, Information & Communication Technology (ICT) has begun to be actively used in agriculture extension. Some of the ICT based channels are:
- Farmers’ Portal: which is a central repository of all the information required by farmers at all stages at one location.
- Kisan Call Centres: where a toll free number 1800-180-1551 is available for farmers to query in regional languages. The system maintains a huge information and receives over 5 lakh calls per month across India.
- mKisan: which is used to send sms advisories by Agriculture Officers and Experts to farmers and other registered users in hindi or English or regional languages.
- KisanSuvidha Mobile App (available on Google Play): which provides all information that a farmer would need from sowing stage to post harvesting including market prices and weather alerts on their mobile. It has been downloaded over 1 lakh times and average user rating is 4.3/5 based on about 2800 ratings.
As per the latest report by The Internet and Mobile Association of India (IAMAI), there are already about 11 crore people in rural India who are accessing internet via their mobiles (not just mobile, but mobile internet!). However, the reality means more than this.
There would be many more rural-to-urban migrants who are working in cities in the unorganized sector but have a family farm back in village cultivated by their family members. Also since an average rural household holds less number of mobiles than an average urban household, we are actually looking at a good percentage of rural households being on internet.
In fact, many farmers have already started creating local groups and started using them to get better terms for their business. For example, the turmeric farmers of Sangli district of Maharashtra already started using Facebook to negotiate better prices for their produce 4 years ago and now have extended their reach to Odisha and Telangana.
The present change is being driven by 3 factors
- Cheaper smartphones
- Cheaper and faster data with wider coverage
- Localization of internet i.e. development of audio/visual content in local languages
With these 3 factors in operation, a lot of farmers will soon be on internet and social media. Our task is simple... to tap them!
We must develop content which is meaningful for them so that they engage with us. If we are able to achieve this, this would be a big step forward in terms of capcity building of our extension mechanism.
This agribot project is one element of the tasks highlighted above. Presently we receive around 5 lakh calls on our call center every month. This shows that our farmers want relevant information today and are willing to ask for it. However, expanding call centers is very expensive and requires huge capacity building.
That is why we must develop intelligent chatbots where farmer can ask in his local language any query and our bot will give a meaningful reply. The project will involve a lot of natural language processing because we can't expect farmers to type commands to our bot (and anyways no one likes to talk to a machine!). So we can't restrict the farmer’s interaction with our bot as “press 1 for this, press 2 for this and so on”. We have to make it as human-like as possible.
Our chatbot will benefit from the localized context information about the farmer. Example, there would be a dedicated chatbot for say paddy growers in Seemandhra or orange growers of Jhalawar which will benefit from the specific knowledge there. For example, it would know what kind of varieties farmers typically grow, what are the diseases they are prone to, what are the solutions, what are the market prices there, what is the weather information and so on. This would also make our task a lot simpler and our bot more intelligent.
The farmer can simply message our chatbot in his natural language and would get an answer. Presently, this thing can be done for facebook but not whatsapp because whatsapp hasn’t opened its libraries yet. Once whatsapp opens its libraries, this project can be taken in the direction of creating local farmers groups. The farmer can message a pre-saved number and will be automatically added to his regional group. Then queries can be made those groups only and also the farmers’ can themselves discuss a lot of other things. Also, with time, we can integrate this with voice recognition and e-marketplace but for now a smart facebook messenger bot should suffice.
Initially we should just look to develop a bot for a specific context to give a proof of concept.