We wanted to make it easier and fun for people to look for rent/sell houses.
What it does
It is a chatbot which recommends houses based on user's descriptions.
How we built it
We have a backend server to which we connect the bot. After user describes a place/house that information is fed into a word2vec model which tries to find the best match from the database, sending the user the respective link from the actual webpage.
Challenges we ran into
Cleaning and preparing the provided dataset is hard. Implementing natural language processing in a real-time application can be challenging as it can be slow to process. Working with different natural language similarity functions to find the best solution for the user's requirements.
Accomplishments that we're proud of
Building a well working chat bot using the Swiss Prime dataset (in German, which we translated into English) using Natural Language Processing.
What we learned
Improving our data preparation skills. Apply natural language processing on a relatively large database. Build a chat bot from scratch. Many different NLP approaches for finding the correct data from the dataset.
What's next for Swiss Prime Chat Bot
Make it even bigger, adding other housing websites from different countries. Selling options. Collect more information from the users before we display our recommendation.