- Intro / agenda
- Leeron: PM on Topics at Google Chrome
- Please see old meeting notes on github
- Topics Taxonomy
- Initial taxonomy includes 349 coarse-grained topics
- Browser classifies site visits into topics from this taxonomy which are shared with callers of the API
- How should the taxonomy evolve over time?
- Josh Karlin (Google Chrome)
- History: take IAB taxonomy and Google’s taxonomy, and find the intersection.
- This is only the initial approach, fully intended to revise or modify
- Principle(s):
- Keep it small to manage privacy, how large is an open question
- In the long run, it need not be something we run from taxonomy, but a general standard
- Other question: What other topics are missing in the taxonomy?
- Brian May (distillery) -
- First bullet under taxonomy -
- Regarding not owning taxonomy: Have you considered IAB’s seller defined audience (SDA) taxonomy?
- It partially covers IAB’s content taxonomy. It is based on the 1p data, it may go beyond content, and include some part of audience taxonomy
- Patrick Soch
- SDA taxonomy included interests, in market and demographic data
- Alex Cone’
- SDA’s value proposition is about user’s inferred interests on a given site/app whereas Topics is about user’s inferred interests across the web
- Josh Karlin
- Purchase intent and taxonomy related to it may be an ephemeral thing and how it will impact - any thoughts?
- David Dabbs
- Chrome will have features that broadly manage product / carts across sites (close to 200?)
- Josh Karlin
- Not sure how the permissions policies will apply for this.
- As for topics is concerned: here is the sites a user is browsing, and what does topics can infer on purchase intent?
- Brian:
- What are the Google Chrome team’s thoughts on this?
- Josh
- We are looking at ways to improve, and if we should revise the taxonomy periodically.
- Brian
- Taxonomy is only one part, but improving and updating it periodically is the other aspect
- Robert Blanck
- Some of the taxonomy questions can be political (across the globe, not just an US centered taxonomy)
- Decision making process and governance - on what is in or out, who can decide and how it can be neutral are some of the hardest questions.
- Brian
- Currently Chrome is making these assignments (topics -> sites)
- How will it be done in future?
- Leeron
- In the short term, iterating the taxonomy for utility and privacy
- In the long term, it needs to be moved to an external governance’
- Michael Kleber
- Is there a proposal for moving governance of taxonomy to a more global approach to account for topics outside of US centered taxonomy
- Robert
- There is no proposal written down, but associations which are reps of the industry will be interested. Google can propose this, and start the discussion, which will make the other partners happier
- Michael
- Is W3C discussion via github and making decisions similar to webstandards the right venue for these?
- Or should we have a different venue who needs to be included?
- Robert:
- W3C is more a technical discussion
- But for the Governance model, and to include others (no technical industry) this needs to be outside. A better support / recognition can be done if there is an external governance structure
- Josh
- What is the target, and what it can accomplish can be done in this room/discussion (what is sensitive, what should be excluded)
- But the overall governance can be done externally
- Achim Schlosser
- Tech and non-technical things can work in parallel
- Brian
- Three sides can work together
- Technical - how many, what
- Non-Technical - What should be included, and excluded
- Middle
- May be a fourth: What would be top topics, and how to rank them
- Three sides can work together
- Ranking top topics
- Leeron:
- Tf-idf, commercial value of topics
- Any thoughts on these?
- Patrick
- Assuming that the top topics are curated by ML models
- Would like to see some inference over time (is the interest informational or purchase intent?)
- Josh
- Purchase intent are also used by other apis (say Fledge)
- Brian
- How it should be ranked likely depend on who we ask
- Potential impact on utility (purchase intent, or ads interest on cpm)
- Alex Cone
- Decision to move from facts to inference is complicated/ huge
- It will also impact the governance model
- Inference may provide utility for some, but less useful for others
- Inference is also what the commercial data providers do
- Brian
- Useful discussion on curating taxonomy
- Alex
- May be discussion can be in dual track; both technical and a governance model
- Robert
- We have been talking about governance for a while, but we have to make progress
- Josh
- Before governance model questions, we are looking into the utility of the api itself.
- Brian
- Once we figured out the technical side, we need to work on their (?)
- Leeron:
- Josh Karlin (Google Chrome)
- Leeron Israel (Google Chrome)
- Brian May (dstillery)
- Achim Schlosser (European netID Foundation)
- Sam Dutton (Google Web Dev Rel)
- Sid Sahoo (Google Chrome)
- Patrick Soch (Merkle | Cardinal Path)
- Jasper Liu (Daily Mail)
- Sathish Manickam (Google Chrome)
- Jen Schutte (P&G)
- Michael Kleber (Google Chrome)
- Christina Ilvento (Google Chrome)
- Maciek Zdanowicz (RTBHouse)
- Tobias Wegmann (BVDW)
- David Dabbs (Epsilon)
- Andreea Cucu (Google Chrome)
- Alex Cone (Coir)
- Thiago Diogo (Google Chrome)
- Trenton Starkey (Google Android)
- Abishai Gray (Google Chrome)
- Daniel Rojas (Google Chrome)
- Marianna Cunha (Google Chrome)