Perplexity Digest

Perplexity Digest is a Django-based dashboard that provides semantic analysis of your browsing history and learning patterns, powered by NLP techniques and the Perplexity Sonar API.

Comment 2

Inspiration

This is mainly inspired by the Perplexity search engine that is up and coming. We wanted to create a project that would align with the usage of the search engine and help readers get insights into their browsing patterns.

What it does

Perplexity Digest analyzes your weekly browsing history to uncover meaningful insights about your learning patterns and knowledge development. It semantically processes web activity in real time, identifies active interests, topics that may be of potential interest and knowledge gaps that can be assessed. There is also a chrome extension which allows seamless integration into browsing habits and develops a simple yet efficient digest for the reader to catch up on at the end of the week.

How we built it

We have used semantic processors for analysing the retrieved search history. Using DBSCAN clustering, we are able to cluster topics of similar interests and in turn display the following using Perplexity's SONAR API:

  1. AI-Generated Insights: Comprehensive learning journey analysis
  2. Learning Recommendations: Personalized suggestions based on browsing patterns
  3. Knowledge Gap Identification: Targeted areas for skill development
  4. Content Summaries: Automated summaries of your browsing sessions
  5. Related Articles: Curated content suggestions based on your interests We have also integrated browser extension found under the subfolder 'extension' in our source code. This can be unpacked as a chrome extension and used to send data to the dashboard.

Challenges we ran into

Our biggest challenge was mostly the formatting of the AI output in a way that readers can easily understand in a concise fashion, hence we made use of semantic processors for our AI outputs. This also means creating a good UI that renders our vision perfectly and we have done our best to give the users a feeling of having an actual digest at their fingertips.

Accomplishments that we're proud of

We are proud of the idea we first came up with and the above challenges that we had to find an appropriate solution to. We are also proud of the UI we built and the way the dashboard itself is organized all in the span of a month.

What we learned

We learnt more about UI concepts, how we can employ Perplexity AI for complex tasks such as the analysis with Browser Topics, which can be a great use case for the Perplexity Search Engine as well.

What's next for Perplexity Digest

We would try to work on more research based functionalities in the app so it accommodates it. This includes:

  1. Suggestions at every stage of the digest so readers/learners know to think outside of what's displayed in the digest(Filling possible knowledge gaps)
  2. Organization of bookmark structures and potentially the retrieval of those topics so that the AI knows more in depth about the interest of the person, so a better curated digest can be developed.
  3. Integration with social media so that each person's digest can be shared and a community space can evolve.

Built With

Share this project:

Updates