Hacker News Reader

Comment by aaronblohowiak | Hacker News Reader

What is the business case for Google in acquiring Aardvark?

One thing they probably really like is the natural language parsing. It's surprisingly good (though by no means perfect) at automatically tagging your question.

For example, looking through my old questions, I think it automatically tagged "what's a good place to get a cheap SSL certificate? I don't think I need wildcard subdomains" as "SSL".

In a perfect world, you ask Google a question just like you'd ask a person a question, and they answer it with the info you want, so they might be interested in the technology and people that make this intelligent tagging work as a step in that direction.

Ah, thanks for that analysis. This is a part that I was missing.

Not to be snarky, but what is impressive about tagging your question with "SSL" - the exact keyword is in the question. No semantic analysis or NLP required - just simple keyword based entity extraction.

I know, it's not brilliant, but SSL wasn't in a particularly special place in the sentence or anything that would make it easy for a computer to figure it out. Compare that question to something like "I am using Rails. How do I ...?", where a computer would have a pretty easy time of figuring out that Rails is important with some simple parsing logic.

Not quite true. Using the standard TF-IDF would have definitely figured out SSL.

Facebook is about getting information from your social circle, while Google search is about getting information from the internet at large. It seems they are looking at various ways to bridge that gap.

It kills a potential rival that could one day grow into a threat. Yahoo didn't acquire Google when they could have. Google has in its DNA the desire to prevent the same embarrassing disaster. IMHO.

Aardvark will integrate extremely naturally into Google Buzz.

When listening to Sergey talk about Google Buzz ten minutes into this video he brings up the value of using buzz to ask questions to his contacts with the recommendation system directing the question to who best can answer it.

http://www.youtube.com/watch?v=d_W6Qbob2mg#t=10m10s

I was like "this sounds like Aardvark".

So why spend $50M when the integration path and implementation are reasonably well flushed out?

Cheaper than a lawsuit later

The recommendation system is not a trivial thing to create, I'm guessing Aardvark's rec system is at least part of it.

http://www.youtube.com/watch?v=d_W6Qbob2mg#t=0m16s

Really?

Pagerank doesnt have anything to do with twitter and facebook.

Its a great idea, and its probably one of the major factors that made google a success, but today it is not a very important factor in google being a bigger company than fb and twitter.

Perhaps I poorly worded by saying "ten minutes in", I meant starting from 10 minute and 10 second mark Sergey brings up how he's been using the service internally.

Edit:But going back to your point, I not sure Pagerank is relevant to twitter/facebook, and I don't think Sergey was saying that it was. But I do think their could be search problem emerging in social networks when people start having several hundred friends and you want to find relevant conversations to your interests going on within your network.

The database could be mined to improve answers in Google. When someone asks "what is the best X in Y", Google will know and MS won't.

They might have some as-yet-undisclosed IP which one of the big search companies would be able to defend - rather own it than fight it.

Also, integrating the already-refined code and ideas into Buzz, Wave, Orkut, other future social initiatives.

Reply on news.ycombinator.com