It can be difficult to handle issues of PTSD, phobias and online harassment when surfing the web since a user's mere noticing the content can cause lasting damage. Here we felt a browser add-on could serve a valuable role by filtering content based on a user's unique triggers or by filtering general harassment.
What it does
The add-on currently runs in one of two modes. The first mode attempts to combat general online harassment using a list of keywords (with respective scores) to determine whether page sub-elements ought be hidden or not. The second mode takes input from a user regarding a 'custom filter' and filters relevant images and text content.
How we built it
We first built functionality that scanned a page and blocked relevant text content. Secondly, we built a tool which analyzed images on the pages on-demand using Clarafai as well as a list of keywords to block user-specified visual content. Finally, we used a Thesaurus API to take the user's initial custom filter keyword and expand it to similar words (say a user blocks 'spider', they might also want to block 'arachnid' and other similar words).
Challenges we ran into
Learning how to create a chrome add-on had a steep learning curve at first since no one in our group was very familiar with the technology at first. Additionally, we found it harder than we anticipated to maintain browser speed while also analyzing the page content.
Accomplishments that we're proud of
We accomplished our initial goal of having a thorough filter tool as well as our stretch goal of analyzing visual material.
What we learned
Most tools, via either help from mentors or online documentation, aren't necessarily as daunting as they might initially seem. Additionally, we learned a lot about Chrome Extensions as well as a small bit about Clarif.ai.
What's next for Rooky - A Chrome Extension for Phobias & PTSD
Flesh out personalization features (allow users to choose how strict the filter is), additionally make the filtering process more visually appealing. Finally, we found that some of the code we wrote would be usable in creating a smarter "Find" feature on a browser (by looking for similar words to the one you're looking for.