CTRLFwiser

Inspiration

Every person working in the office spends long hours looking for information. That time could be spent on doing actual, productive work. If we could just make looking for information more efficient!

PRESENTATION: https://drive.google.com/file/d/18LosBGzHiPN1HDXbxl7FDvLdtQQC6XgL/view?usp=sharing

What it does

It uses ML to match the search results not only using keywords but also understands the intent, context, synonyms and doesn’t care about your typos. It can mark the most relevant text in the document or part of the system/application for your question. It can also show you alternative answers, matched with less probability but also relevant for the specific question. Depends on the matching accuracy, it marks the text with low or higher transparency so we can easily identify all results.

How we built it

Developed on a hackathon: Engine is based on word2vec word embedding and allows us to find words that are similar in the linguistic context.

Challenges we ran into

We have never developed Chrome extension and we had challenges with the quality of ML. After the research of the last state-of-the-art NLP, we managed to create a working solution.

Accomplishments that we're proud of

We are proud that during just 24h we manage to build a plug-in which is working with Google Chrome and already makes a better use from ctrl+f

What we learned

How to develop a chrome extension and how we can use different ML models for particular use cases.

What's next for CTRLFwiser

We plan to develop it into the intelligent knowledge base which combines multiple sources of information (cloud drives, slack, e-mail, intranet site, wiki) and you can ask natural language questions and get answers instantly, directly from your website.