Selected answers from the Dumb SEO Questions G+ community.
Casey Markee: “Although many NLP systems are moving toward entity-based processing, most still identify important phrases using classical keyword-based approaches. To bridge this gap, we introduce the task of entity salience: assigning a relevance score to each entity in a document. We demonstrate how a labeled corpus for the task can be automatically generated from a corpus of documents and accompanying abstracts. We then show how a classifier with features derived from a standard NLP pipeline outperforms a strong baseline by 34%.”http://static.googleusercontent.com/.../archive/42235.pdf
Casey Markee: Melissa, yeah, no. Certainly more to it than that. But understanding it and using it now is still a big deal.
Melissa Fach: It is hard to explain to people without clear examples, why things are not so simple. You can`t recommend things without explaining how it works.
Bill Slawski: The process within the paper is a little dated because Entities identified in Freebase has been replaced by Wikidata IDs. Those are mentioned in this part of the paper:"Given a document and abstract, we run a standardNLP pipeline on both. This includes a POStagger and dependency parser, comparable in accuracyto the current Stanford dependency parser(Klein and Manning, 2003); an NP extractor thatuses POS tags and dependency edges to identifya set of entity mentions; a coreference resolver, comparable to that of Haghighi and Klein, (2009)for clustering mentions; and an entity resolver thatlinks entities to Freebase profiles.
Michael Martinez: You can read the original paper here: http://www.aclweb.org/anthology/E14-4040
Michael Martinez: The paper explains how they trained an algorithm to predict how important an "entity" would prove to be, given enough documents that mention it. Since they were using abstracts there is no way this could be leveraged for search engine optimization for the general Web.
Michael Martinez: I cannot see where Daniel Gillick has published anything relevant to this work after 2015. Looks like he moved on. I will link to a few things in replies to this message.
Travis Bailey: Okay, now that I revisited this tab two hours later - I think I was being a little bit of a smart ass. Just look up the definitions of `entity` and `salience`. It literally boils down to what we were, or should be, doing anyway.Say you`re selling kittens. You want to be a badass kitten trafficker. So you plan out the site... like ya do. Slam it together, and start getting relevant. But how is relevance established? When have you hit `top o` the mountain` in relevance? What`s for lunch?First, ya get cozy with all things related to kittens. Rub some elbows with people/sites, firmly ensconced in all kinds of things kitten related. Hopefully you`ll start getting links and mentions, and (GASP!) maybe get to know-ish a few people.Taking that notoriety, you work your way up the food chain, Elliot Ness style. Eventually your business/site may be rubbing elbows with notables such as Saul Kaplinksi (The Kitty Litter King of Tuscaloosa), and Jeana Rawley (Cat Toy Maven out of Charleston). If you`ve done a really good job, you get links and mentions from Kitten Trafficker Monthly. Much success!So what they`re talking about is the reverse of that. Connecting the dots from the top down. And hopefully it all results in a good SERP for kitten related queries. Yep, they`re just looking for a good way to establish who is an actual authority - page by page. The idea is a science-y jargon treatise on what we call Authority.Anyway, I kind of want a grilled cheese for lunch. Thinking cheddar jack and muenster... but I`m all out of muenster. :(