While studying for finals, we realized that it would be useful to revisit portions of recorded lectures that cover specific topics. Often these topics were visited at many points throughout the course of the semester, so it would be difficult to find the relevant lectures simply based on lecture dates.

What it does

Based on the above problem, we decided to create a platform for searching for relevant timestamps in lectures based on keywords. After searching keywords, the user is shown a lecture that starts playing at the most relevant timestamp. Additionally, the user can view other relevant timestamps and lectures. To enhance the studying experience, we also created a search bar on the side of the UI which displays related content from academic websites that could provide background information.

How we built it

This project uses google cloud video intelligence to extract transcripts from the videos. Keywords are extracted from these transcripts using the rake open source api. These keywords are stored along with the corresponding timestamps and video data in the google cloud datastore. When keywords are searched, the datastore is queried for the searched phrase as well as words in the phrase. The best matches are chosen from the datastore response. The top match is displayed, and the other matches are listed beneath the video. In the side bar, related search results are displayed from a bing custom search that only searches through sources that primarily contain background information.

Accomplishments that we're proud of

The user interface with the video player, background information search bar, and related videos creates a very functional and productive study hub.

What we learned

As our first hackathon, we witnessed the time crunch and challenges needed to deliver in such a short window. We learned how to divide tasks with clear specifications to ensure that the disjoint tasks could work together to create a cohesive final product.

What's next for The Last Lecture

To expand The Last Lecture, we would add more videos and continue improving the algorithm for choosing best matches.

Built With

  • azure
  • bing-search-portal
  • datastore
  • gcp
  • videointelligence
Share this project: