Alexa and RPA based College Information Provider
The Oriental proverb quite aptly states,
“Choose a job you love and you will never have to work a day in your life.”
Easier said than done, right?
Fresh out of high school, even if you have all your interests figured out, it is an overwhelming endeavour to sift through the myriad of subjective information regarding the options to advance your education in the way that fits your needs.
Because the education sector in Jammu and Kashmir is not exactly streamlined, the available information about its component institutes is in absolute disarray: in bits and pieces, all over the internet. Sifting through that information might be more of an exercise in futility, and less of a career guidance trip. With this information gap in mind, a system that streamlines this information in an intuitive and intelligible form is need of the hour.
ALEXA BASED COLLEGE INFORMATION PROVIDER is an attempt at developing such a system.
I know it sounds like a mouthful but bear with me for the next few minutes.
What it actually does:
The basic model is intuitive in its execution. A user queries about colleges or particular courses available. The user does so in a voice-driven interactive manner with Amazon’s virtual assistant — Alexa. The model prompts a college or a course recommendation alongside other information like prerequisites, intake capacity for the course.
What goes on in the background:
Suppose you were to ask Alexa the following query :
“Alexa open j. k. e. d. u. and tell me about courses provided at Model Institute of Engineering and Technology”.
The Alexa enabled device picks up this message and forwards it to Alexa Servers.
A speech engine records this sound and represents it in a format that enables it to recognize individual characters spoken. This is usually done by feeding small audio slices into a neural network, which in turn tries to predict the letter with which it corresponds. The combination of these letters along with likelihood scores based on large corpuses, a transcription of the message is generated. The key words are identified to make sense of the query. This information is then used to map the query to one of the Amazon Skills provided by the Amazon Skill Set.
The Amazon Skills Kit provides a more robust Alexa beyond generic commands. Skills are essentially apps on the Alexa platform. A framework of Skills based around our app maps specific queries to responses as computed by the information provider. This works by matching the given information from speech to predefined text patterns and when a match is found, the accompanying intent or trigger is executed. In our user query, lets assume it is matched to “courses-query” intent.
The trigger or intent is forward to Amazon Lambda Functions. Lambda Functions enables us to define API endpoints which are triggered by Alexa Skill Kit. Amazon Skill Kit makes an HTTP request to the intent corresponding Lambda Function. The Lambda Function in this user query will build up a response for the user using the data from the database in the form of a string. This string is sent back to the Amazon Skill Kit as an HTTP response.
The HTTP response from the Lambda Function is then forwarded back to the user device from the Alexa servers where the Alexa text-to-speech service reads out the message “Model Institute of Engineering and Technology offers engineering…”. Thus, completing the query.
It is essential for the functionality of our application that the information in our database is refreshed periodically. To do this efficiently, an RPA web scraper and orchestrator is set up. The system uses UiPath web scraping to scrap the websites and gather the latest information about the institutions. This information is then stored in the MySQL database which is later used by the Lambda Function to build up the response. Adding to this, a web interface build using HTML, CSS, PHP and AJAX, also accompanies the system as a visual medium for the users to browse through.The web interface and the MySQL database is hosted on Amazon Web Services(AWS) EC2 instance. There is a PHP script in the backend that exports the MySQL database into JSON file. The PHP script also has a cronjob which runs, producing the updated JSON file, whenever there is a change in the MySQL database. Alexa skill’s Lambda Function uses Remote JSON API with the JSON file link as the parameter to parse the JSON file. Lambda function parses the Json file in response to different intents and their respective utterances to answer the query put forward by the user.
