Parse PDF using the browser's built-in AI function

Parsing PDF? Only the front end is enough！In a weak network or restricted offline environment, you can translate and summarize PDF.

Comment

Inspiration

Develop a document system that users can use to view PDF files. It is forbidden to access the external network. To realize text-to-speech, summary, and translation, the backend needs to develop corresponding services. It would be great if you can just use the Chrome browser to implement it offline.

What it does

Summarization and translation can be realized with just Chrome support

How we built it

First parse the PDF document, and spell out the parsed fragmented and incoherent sentences into correct sentences that can be read by humans according to the language characteristics

Challenges we ran into

  1. What can users do when tasks that take a long time are in progress
  2. How to deal with the text limit that exceeds the summary task
  3. Optimize the text algorithm as much as possible in rendering
  4. Complex data interaction

Accomplishments that we're proud of

  1. When executing long-time tasks, users can switch to view other files freely. Before the previous task is completed, new tasks cannot be executed
  2. When the task is completed, there will be a notification to tell the user that you can switch to view the results or perform a new task
  3. Break the text into units of about 3,000 words

What we learned

Chrome Built-in-AI can allow front-end developers to further contact The AI ​​world has expanded new front-end areas and new depths. This is my first time to participate in devpost challenge, especially for my favorite company Google. I am very happy. As the end date approaches, I am running at high speed like a productivity machine.

What's next for Parse PDF using the browser's built-in AI function

I spent a long time thinking about what to do, and it took about a week to actually start working on it. If I have enough time, I will complete the corresponding Chinese-English translation function, generate summaries based on the translation, provide a better interactive experience, etc. I hope that Google can open text-to-audio and image-to-text interfaces in the future to release productivity.

Built With

Share this project:

Updates

Submission history