We’re back in action here in Santa Clara! It’s time to talk about real-time search and how the engines are handling it. Will it ever stop enabling real-time spam. What’s the deal and what are the engines doing to make it useful? How can marketers get their content to show up? That’s what we’re gonna talk about here.
Up on stage we have Danny Sullivan moderating speakers Dylan Casey, Ivan Davtchev, and Sean Suchter. In the peanut gallery you have a liveblogging shivering. Just saying. It feels like Troy in here.
Up first is Sean Suchter.
Why do we all care about real-time? Because users expect to know right now. They want the answers to:
- Did I just feel an earthquake (YES, California, YOU DID. Stop tweeting it. This is how the dinosaurs went extinct.)
- What’s going on with the Giant’s game? (I hope they’re losing)
- Is my cell network down? (Is it AT&T? Then, yes.)
What are the challenges to keeping up with real-time search:
- Keeping up with speed
- Making sense of massive content
- Keeping it relevant.
Twitter content is timely, but it’s not always accurate. It’s often made up of opinions, reactions, impressions, etc. It doesn’t come with context. You need to know who said it to whom.
A few weeks ago there was a plane crash in Palo Alto. One of their engineers was trying to figure out what happened. He searched the traditional results from Yahoo and he couldn’t find anything. It just wasn’t in there yet. Then he tried Bing Twitter and it was right smack there. But you had to search social specifically in order to find it. With real-time search you’re fighting all the different public platforms and whether they’re public, not public and where partnerships exist.
To rise above the noise, there is more to do as search gets more social.
New things to consider:
- People retweeeting what you said, share good links
- Trusty-worthy people sharing your links and tweets [So…the search engines are now determining how important you are? How’s that working? Is it based on follower counts? Cause, that’s not gamed like a sex pill.]
- Number of people retweeting what you said in the last minute, hour or day
- Be prepared to turn on a dime.
Real-time relevance needs data mining to filter and rank based on history. Spamming communities can be highly visible. He finishes up by looking at Bing Twitter. It shows the top tweets, the top shared links, and the top tweets/sentiment per link. Sweet. For the seven people using Bing. I kid, I kid!
Next up is Ivan Davtchev.
What is real time search? It’s showing the most relevant up-to-date content for a topic of recently increased interest. The freshest content is great, but it’s not always the best. Yahoo puts Twitter in its news page and in regular results.
Challenges of Real-Time Search
- Real-time indexing: Get new content as its published. You have to crawl really, really fast.
- Query analysis:You also have to determine which queries you want real-time search showing up for.
- Ranking for fresh content: it doesn’t have proper anchor text. How do you find it?
How do you know a query is really hot?
Find, in real-time, queries about emerging events and news stories. He talks about Google Trends as a way to show how popular a query is. It works well for head queries, but not so much tail queries. It’s the standard approach, but not ideal.
They have Yahoo TimeSense (it’s not public, sorry). They use it to detect what’s spiking. That’s how they find out what’s trending, what Twitter says, etc. It’s different and better because it uses language modeling. Language is a collection of words, it’s a way to explain vocabulary to a computer.
They build small language models for slices of time. Is this query much more prominent right now? They translate the question to something like, “is this query more related to the last X minutes or the next X minutes?” It lets them capture related queries that use the same vocabulary of spiking terms.
There is more secret sauce of course.
- Perhaps building language models for fresh content like Yahoo News and Twitter.
- And doing this very fast in production.
- And then ranking content in real-time.
Yahoo Real-Time features to look forward to
- Ranking + indexing even closer to true real time
- Real time results in search verticals beyond news
- Real time relevance algorithms powering experiences in Yahoo! properties beyond Search
Real Time Practices To Avoid
- Do not create content with multiple buzzing terms. That looks like spam. [I don’t know what he’s talking about. miley cyrus justin bieber glee.]
- Do not abuse shortening services for spam links.
- Do not go overboard with Twitter hashtags. It’s not visually appealing. [YOU CAN”T TELL ME WHAT TO DO! #hashtagsarepeopletoo #ilovehashtags #hashtagsarethebestpartoftwitter]
Next up is Dylan Casey.
Why is real-time information important? All of these tools (social networks, chat, Twitter, etc) have turned anyone into a publisher. As the publishing barrier became lower and lower, the necessity for speed and relevancy was going up. We had blogs, we had better publishing tools, YouTube, etc. They thought about how to look at the content. Not only is the content different but people are using it differently.
Why the explosion of information? Because it’s so easy to send out a Twitter update. It allows you to speak much differently. It’s like sending a text message vs an email. They wanted to build something that was part of Google. They didn’t want to create a standalone product. [Take that, Bing Twitter!] People are already using Google. [I guess that’s why Bing creating a standalone product…] Real-time search is like pointing a web cam at what people are searching for right now.
They launched real-time search results in December. They’re adding new providers like MySpace, Facebook and Google Buzz. He does a demo of real-time search. You know, in case you haven’t touched a computer in three months. You search for Toyota and there’s scrolling real-time results. You can go directly to a real-time Toyota page.
Since they were relying on Google Web search, they weren’t able to leverage all the technology that they use to rank documents. They had to come up with something new. They look at the number of followers. They look at author quality. They needed to be careful when deciding whether or not to show real-time results into the SERP. They’ve had huge debates trying to understand whether something was a useful answer.
They want to create an open platform for all types of information, not just the large providers.
How do you get into real-time search? Find something you’re passionate about. Write often. Don’t overdue SEO. [omg shut up. Those are the super secret recommendations? PFFT!]
And that’s it. Time for lunch. We’ll be back later. Go grab yourselves a snack at home.
