Data science in Media: Creating and assessing media article embeddings

Register Learn more

The December edition meet-up from the Data science society meet-ups 2022 is under the topic of Media!

Who is the presenter?

Krum Arnaudov – Data scientist @ Financial times – is working on NLP tasks, subscriber churn and lifetime value.  Before FT, he has worked on ML projects in the healthcare industry – insurance claim classification, in/outpatient time series prediction, claims novelty detection. He feels equally comfortable with Python and R, loving the flow of a R EDA-to-modelling with the latest Rstudio stack (tidyverse + tidymodels), while also enjoying the maturity of the Python ML-verse. 

Before falling in love with DS/ML, he used to do B2B Key Account Management and Ops Management.

Outside of work, he enjoy spending time with his wife and two daughters, making music and following the local politics (truly!). 


During the event you will have a chance to learn about creating document-level embeddings for a long documents – an ongoing topic of research in NLP. We will:

  • present the lessons learned from embedding media articles at FT
  • compare pros and cons of methods such as TF-IDF, pooled Word Embeddings, Doc2Vec and SentenceTransformers
  • how you can approach the assessment of the results. 

Except for the MeetUp, everyone is invited to share what s/he is doing in the field of Data Science – We believe that sharing is caring! Save some time after the presentation for networking and chatting! : )

Data and Time: 13.12.2022 , 19:00-20:30 EEST Timezone

Location: Financial Times Sofia , Moskovska Street 9


The event is open and free . Book a ticket to reserve a place and to get the materials afterwards.

Prepare your questions for the speaker and ask in the Q&A part from the event!


Event Partners

This event is organized in partnership with Financial Times. We want to thank them for the support and their contribution to the Data Science Society!

Bookings

Bookings are closed for this event.