The Coming “Entity Search” Revolution (#smx #24C)
It’s the end of the day, we have one more session to go before the SMX After Dark networking extravaganza, and I’m still in the Semantic track – which means I am in a pretty darn good mood.
Danny Sullivan (@dannysullivan) — the session moderator — starts the session by saying how he hates the word “entity” nearly as much as he hates the word “semantic.” They’re two words that mean so much but express so little, he says.
He goes on to say, despite hating the word “entity” as a vague umbrella word, the actual concept of “Entity Search” is really opening up a whole world of possibility – a world where search engines don’t just match words with words, but actually begin to understand ideas.
An Introduction to Entities
David Amerland (Author, Speaker, Analyst, @DavidAmerland) starts the session with an introduction to entities. David is going to talk about entities, how they apply, and what you can do to take advantage of them.
He asks how many people know what an entity is (by show of hands). I think more people know what it is than raise their hands (we’re all scared of David’s intelligence). He asks us next (by show of hands) how many people know what a tree is. Again… more people should have raised their hands (who doesn’t know what a tree is??). He coerces us to raise our hands (we all know what trees are!) and then he asks us to close our fists and close our eyes.
I am explaining all of this because it’s a great demo.
With our eyes closed he says “tree” and notes that we all have an image in our heads right now. Without prompting, we all see different tress based on our understanding of what a “tree” is. If we know more than one language, when we hear tree in German we see a tree. If we don’t know German, we don’t see a tree because we don’t know that Baum means tree.
Everything you know about a tree is your own personal knowledge graph. You have your own personal contexts associated with that word.
Someone who has grown up in the forest sees a different tree than someone who grew up in California
Trees are entities. Entities are concepts. We (and Google) understand entities (Concepts) based on context, experience, influence, social factors, societal factors, etc.
David jokes that as SEO Professionals we used to know how the web works. “We used to understand the Matrix” he jokes.
SEO in the past was a case of probabilities; search was a statistical analysis tool. Search returned keyword-rich results.
As SEOs we’re responsible for understanding the mechanics and delivering the results. As semantic web statistics begin to fade away, search itself is changing.
A good example of this is Google Now. Google Now is actually learning what you need, and it’s trying to predict what you might need. For instance, if you look up directions to a place, the next time you use your phone it might let you know that the place you looked for (IE: The convention center you visited yesterday) is nine miles away and traffic is bad.
The challenge for us is to understand the new landscape and what we can do within it. Information itself is growing at an exponential rate and search is changing with it.
Vectors of Big Data
Volume – how much you create
Velocity – how quickly it accelerates across the web
Variety – are you creating a good mix?
Veracity – make sure what you’re creating has a large element of trusted authority (directly related to Authorship)
Authorship + Author Rank
Entities are produced in a variety of ways
Imported from trusted sources
Extracted rom web pages
Data mined in the social web
There is a massive amount of the web that is unstructured data. Google’s job is to take this unstructured data and put it into a structured format.
A tree (re: the entity) is a defined concept that is defined by everything it is associated with.
Entities are a challenge because they provide answers at the search box rather than suggestions.
“Optimization” seems to have been reduced to this single problem: Find a way to answer a single search query.
Entity Search, and Google’s emphasis on it isn’t all bad! It’s created a lot of opportunity, actually. We need to ask ourselves:
- How can we learn to adapt to optimization for continuing search interfaces (touchscreens, etc.)
- How can we be less reliant on keywords
- How can we optimize in a world of deprecating on-site SEO
Reverse engineer entity as a driver by thinking about your content as needing:
Web used to be faceless, impersonal, able to be gamed. Web is now personalized; everything you do is associated with and accrued to who you are and what you do.
What is Freebase?
Metaweb ten acquired by Google; over 40 million topics; over 2k relational tables; had a metadata query language; you (can) use it to build applications.
Here are some slides that explain what Freebase is better than I can:
Freebase query editor
The future of search = Semantic + implied intent. Your personal historical previous searches, purchases, and sphere of influence all directly affect SERP results.
It’s going to get a lot harder. We all need to try to build a sphere of influence, and we need to start marking up our pages/content with Semantic Data.
Jon predicts “in 2015 backlinks will mean nothing.” He’s not sure if it will happen, but he’s hopeful.
She is going to talk conceptually about how people become things, or – in other words – how the Knowledge Graph becomes the Social Graph.
She opens up with a few quotes. One from SE Land and this one from Matt Cutts: “All these people doing [… ] over optimization – versus those making great content and great sites. We are trying to make GoogleBot Smarter, make our relevance better…”
She notes how this “over optimization” technique represents the old, traditional world of SEO and SEO techniques, and the latter half reflects where search is going (IE: toward entity search).
SEO and SERPs today are all about more data, not less. We’ve evolved from text and links and files to things. We’re able to (and consistently try to) make our communication more comprehensive.
Google is relying on the social graph, the knowledge graph, and links.
Google determines your preferences (and your personal search results) based on your interactions with others on the web and how they interact with you.
SERPs are showing more content endorsed by your friends. How do you show up here? Presence on G+, Authorship and engaging in these spaces
Rich Snippets make flat pieces of text become things. How a person becomes a thing is through G+ verification.
She emphasizes the importance of AuthorRank. She thinks of it like PageRank. It’s really about making connections, being authentic and participating in the community. (IE: Being a player in your “sphere of influence” as Jon put it).
She says SEOs and Social Media Marketers really need to work together and reiterates that one key takeaway should be to markup your website with as many (relevant) schemas as possible.