-
Notifications
You must be signed in to change notification settings - Fork 0
Community Bonding Period
During this period I took a meetup, organized by the org admin, with all the mentors/contributors for Google Summer of Code 2023 at DBpedia. In this meeting, some interesting statistics was shared with us, like the percentage of successful previous contributions, percentage of contributions that became a paper in collaboration with the mentors, number of project proposals sent for GSoC 2023 and number of accepted ones.
This meeting was also a opportunity to present myself and share my project with all the other contributors. And also, a opportunity to ask some questions.
In another meeting, I also met my mentors for the first time. During this meeting, my mentors shared some useful information about the project, from technical tips to personal advices of how to make the best of this experience. I fell relieved by the fact that the project is flexible regarding its deadline and I don't need to worry about not delivering everything as planned for the week when my schedule is jammed with external responsibilities. During this meeting, we also aligned the start of the project. We defined the most important RS's datasets to integrate with DBpedia, considering their popularity and the data available by DBpedia.
It was great to bond with the community. Nice to meet you guys! Looking forward to collaborate with you all! Feel free to reach me out, about the project and for any help.
I also took the community bonding period to warm-up for the project. I used this time to learn more about DBpedia, Recommender Systems, Open Source standards and ML frameworks for reproducibility.
I also experimented the baseline for integrating the MovieLens dataset with DBpedia. In this notebook I matched some movies with their respective DBpedia's URI using their name and release year. With this simple baseline I was able to reach a coverage of about 64% entities of the dataset. Now, I'm seeking for other ways to improve this baseline. While reviewing the literature, I was able to find some other works that already entity linked some RS's datasets with DBpedia, but I was unable to find in full details how the match was done.