ΕΕΛΛΑΚ - Λίστες Ταχυδρομείου

Re: Contribution to PersonalAIs

Dear Yan,

Thank you for your email and for sharing your background with us. It's
great to hear about your interest in the PersonalAIs project, and your
experience in Music Emotion Recognition, NLP, and Full-Stack Development
aligns perfectly with the goals of the project.

The PersonalAIs project aims to develop an AI-powered agent for
personalized music recommendations by leveraging natural language
processing (NLP) to understand the user's emotional state and musical
preferences. The system will generate and refine music playlists based on
user input, utilizing the Spotify API for authentication and playlist
management. It will also support real-time conversational modifications to
playlists, allowing users to adjust mood, genre, and energy preferences
during their interaction with the agent.

Core technologies include:

   - NLP techniques for detecting user mood and preferences
   - Generative AI models (small LLMs) for dialogue generation
   - Integration with the Spotify API for user authentication and playlist
   management
   - A web-based chatbot UI for interacting with the agent
   - Real-time playlist modifications based on user feedback

In terms of your inquiry about evaluation metrics, we are considering using
both industry-standard recommendation evaluation metrics like NDCG
(Normalized Discounted Cumulative Gain) and MRR (Mean Reciprocal Rank) and
a more user-feedback-driven approach. A hybrid approach could help us
balance objective performance metrics with subjective user satisfaction,
and we would welcome your thoughts on how to best combine these methods.

Looking forward to your continued interest and contributions to the
project, and please feel free to share any further questions or ideas you
may have.

Best regards,

Giannis Prokopiou & Thanos Aidinis

Στις Δευ 17 Μαρ 2025 στις 6:41 π.μ., ο/η Yan Zhang <akira [ dot ] zhangy [ at ] gmail [ dot ] com>
έγραψε:

> Dear Giannis & Thanos,
>
> I hope you are having a good day. I am Yan Zhang, a master student in IT &
> Cognition at Copenhagen University. I have been closely following the
> discussions around the PersonalAIs project, and I am very interested in
> contributing, and it aligns closely with my background.
>
> With my background in Music Emotion Recognition, NLP, and Full-Stack
> Development, I have worked on several projects that align with the goals of
> PersonalAIs:
>
>    - Music Emotion Recognition: I conducted a comparative study on the
>    performance of unimodal (audio or lyrics) and multimodal (combined audio
>    and lyrics) models in Music Emotion Recognition (MER).
>    - NLP & Machine Learning: I have experience in text classification,
>    sentiment analysis, and author style recognition, using a machine
>    model with NLP techniques.
>    - Full-Stack Development: I've developed a responsive e-commerce
>    platform for yoga products, featuring real-time price calculations, user
>    authentication, and secure payment processing. Built RESTful APIs using
>    Express.js, designed the frontend with React.js, integrated Stripe API for
>    payments, and implemented MongoDB for data storage.
>    - Co-Pilot and Question Answering System – GraphRAG for Workflow
>    Understanding: I'm currently doing my thesis about GraphRAG and fine-tuning
>    SLM models, which could be useful in exploring how PersonalAIs can leverage
>    models for interactive recommendations.
>
> By integrating these experiences, I believe I can contribute effectively
> to developing a personalized music recommendation system for the
> PersonalAIs project as part of GSOC 2025. I am excited about this
> opportunity and eager to explore how I can contribute.
>
> Additionally, I would like to ask about the evaluation criteria for the
> project:
>
>    - Does PersonalAIs have predefined evaluation standards? If not, do
>    you think it is necessary to establish them?
>    - Would you prefer to use industry-standard recommendation evaluation
>    metrics (e.g., NDCG, MRR), or would a more user-feedback-driven approach be
>    more suitable?
>
> I would greatly appreciate your insights on this. Looking forward to your
> feedback!
>
> Best regards,
> Yan Zhang
> MSc IT & Cognition, University of Copenhagen
> ----
> Λαμβάνετε αυτό το μήνυμα απο την λίστα: Λίστα αλληλογραφίας και συζητήσεων
> που απευθύνεται σε φοιτητές developers \& mentors έργων του Google Summer
> of Code - A discussion list for student developers and mentors of Google
> Summer of Code projects.,
> https://lists.ellak.gr/gsoc-developers/listinfo.html
> Μπορείτε να απεγγραφείτε από τη λίστα στέλνοντας κενό μήνυμα ηλ.
> ταχυδρομείου στη διεύθυνση <gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr>.
>
----
Λαμβάνετε αυτό το μήνυμα απο την λίστα: Λίστα αλληλογραφίας και συζητήσεων που απευθύνεται σε φοιτητές developers \& mentors έργων του Google Summer of Code - A discussion list for student developers and mentors of Google Summer of Code projects.,
https://lists.ellak.gr/gsoc-developers/listinfo.html
Μπορείτε να απεγγραφείτε από τη λίστα στέλνοντας κενό μήνυμα ηλ. ταχυδρομείου στη διεύθυνση <gsoc-developers+unsubscribe [ at ] ellak [ dot ] gr>.

αναφορές

πλοήγηση μηνυμάτων