I am not a live blogger, but wanted to furiously type notes for you all that could not be here. Hopefully you will get a nugget out of the scramble of notes below.
Beyond Algorithms: Search and the Semantic Web
New faceted search engines are emerging that promise smarter and more personalized results that take advantage of the Semantic Web. Do they deliver, and what do these engines mean for traditional search? How can obstacles such as scalability and diverse content provisioning platforms be overcome for Semantic Search to succeed?
#beyondalgorithms (Comically Danny adds, “you may have some room after that tag to actually tweet”)
Gil Ebaz: Founder and CEO of Factual – works in sharing and mashing structured data. Simplifying access to data through community based tools.
Will Hunsacker: CEO of Every. Business.com, Goto.com and Overture.com. Consumer site searching web for most interesting, timely relevant news.
Barney Pell: Whizbang Labs tried to treat web as an entire database, founder CEO of PowerSet – natural search engine acquired by Microsoft in 2008, now works with Bing matching words in documents to help solve complex tasks. Thinking about the meaning of information.
Nova: Works with a “do” engine, creating a virtual assistant. Can download to iphone and soon to Android.
Barak Berkowitz of wolfram-alpha – doesn’t know what semantic web is. What is wolfram-alpha? 10 trillion data points that uses parser and interpretive engine. Ask Wolfram Alpha a question, “how many calories are in cup of chicken, cup of peas”… it calculates it exactly. Overcoming need to know how to get that data, mash it up and give an answer to a question that is relevant to you.
Moderator (Can’t find his name, but said, “I’m the moderator” when Danny was making a point) Last five years with Twine working on what is where. What is on a page and what does it mean, now interested in time dimension. Live Matrix, what is happening when and where on the web like TV guide for what is happening on the web. Semantics will play a role.
Carla: VP of Community Strategy at Guidewire, focused exclusively on start ups.
Danny Sullivan: Remembers life before Google, from Search Engine Land. Semantic web definitions so bad, so misunderstood, doesn’t know that semantic is the right term, the web is a messy adversarial place where people are trying to trick search products making jobs those trying to organize data harder.
What is Semantics:
Barney: About meaning. Literally. Literal matches and literal features. Great technology applies to everything from source code to DNA. Abstract features, not tied to how it is expressed, lever of abstraction that has meaning to people.
Semantics is middle layer connecting user behavior to higher level of actual intent that users have. Contends Google and Bing are already semantic search engines.
No one is going to search engine to find semantics. Semantics is a technology term, not equivalent to search.
Carla: Confusion, when discussing “its all geek speak”, now gets eyerolls when mentions semantic. However, when show technology that uses semantics it brings it together.
Gil: Offers consumer more features if developers use it in the right way. How computable is the information available. Increasing structured data sources that might become accessible, crawlable.
Berkowitz: Not a good description of everything. More information out there than you can get from a search engine. Different ways to deal: use different approaches to understand meaning, mine data that is computable data. If all this knowledge is available to people, semantics can help people discover more information.
Nova: More interesting is what is happening now is computers are starting to understand what people are asking them to do. My computer can begin to understand what I want it to do. When asks for flighttime and understands it is running late, auto checking for avail at hotel closest to airport. Can help solve real problems, notable technology can enable it.
Amazed by how little creativity there is in going beyond search. Lack of creativity in determining relevance, vs having biggest database. How do you solve people’s problems vs who has the best blue links.
Danny: Semantics – you have to understand the meaning of something, which is why “semantics” term should be killed as no one can define it. How people link is semantic search, Excite used to say they did semantic search, if search for apple you might be interested in orange.
Carla: Next phase of semantic search is about presentation and aggregation. “Google works pretty well for me” Way to look is how the results and data are presented to me in a way that really works for me. The Google result list is horrendous, answer my questions in a way that makes sense. Subset of data geeks, looking at all the data, how to aggregate and use it. More interesting than natural language.
Berkowitz: Opportunity to give an exact, single answer to what you are looking for. Challenge is presentation of results. All at once? One pure answer, then more results that may be related. Ideal thing is if you ask the question when is high tide in Monaco April 30th, you can get exact answer. Just scratching surface with 10trillion data points.
Danny: Search is not outdated. It works, because a good way for people to get data. Is happy with the way search works, so are others otherwise they would be out of business. Suggest that the whole thing gets thrown out, because it is not in fashion. However, different things make sense at different times, maps are a major revolution, pivot tables Microsoft, Google squared factual, wolfram.
“heaven help us if every search on Google turns into Wolfram-Alpha”
Moderator: Solutions that worked don’t necessarily scale, at some point users give up. What is the point in having the other pages in results.
I don’t know an engine that does not try to understand the meaning of what you are searching for.
Berkowitz: When people are cre
Nova: Search and results in conversations tells him have not escaped paradigm. Data retrieval vs. “do something for me” goes beyond finding something on the web. It is not just about information retrieval. Do something new.
Barney: When you look at what people are doing with search engines, when you don’t get the result in a complex task. Many sessions last more than a half hour for complexity. What movie is showing, where can I park…
When think about broad perspective actually solving task…..are people completing tasks? How many queries to complete them. Wants system, to get out of the search box.
Nova: Context matters. Know who is asking, personalization and context have important role. If you could just go and ask a question, “do this for me” “take care of this for me”, some news there….
Berkowitz: API for search that can be used in different places. Ability to mash things together.
Gil: What innovation is there? Decision trees, like hunch, Microsoft pivot, seer. Possibility for start ups to tap into data
Will Allen: Real-time web. Searched for search engine in Google, #1 result was Alta Vista. Find important information that is happening now.
Want people to search less and have information delivered to them proactively. I don’t think consumers care what is happening under hood, use natural language processing to do extraction in building structured database on people, places things. Understanding the intersections, who is doing what to whom? Can actually measure sentiment.
Danny:
Search is the mall, Google is the anchor store, doesn’t mean they cant all grow, but difficult to unseat that. Is there an app for that? Is there a Google for that? Search activity providing you an answer. Many activities on your mobile are search activity, but done in different ways vs regular query. Doesn’t occur to him to open Google on mobile to find a restaurant, example that Google does not, and will not own it all as they (foursquare etc) are growing.
No more normal results in Google since December, logged in or not your search is personalized based on your cookie unless you search without cookies or block ip, not very many do.
Berkowitz: Most critical question: Who is asking for the information.
Carla: we have to be more public with our data.
Question: Works in higher education, wants to improve delivery of courses to adults online. Is technology coming that can tell if student clicks here and tell how long it takes a person to find answer.
Question: Where is trust in information retrieval?
Danny: Not just number of links, assign authority to every website, once again, “if trusted site you can fart and rank”. Google can tell when you came back, session is created. Reference to Bing search sessions that go on for two months. Tracking goes along to ad serving and more.
Nova: Transparency, make it easy for the user to determine how much they are giving.
Odds and Ends:
Danny: Facebook search is terrible, it will not replace Google. Facebook search is terrible for searching facebook itself.
Danny: Growing concern for SEO’s where data in descriptions comes from. As engines provide more answers, provide support for web content producers if traffic goes away, yet engines still benefit then they stand to be viewed as leeches.
Search Engines are largely an un-superficial bunch. While the way your website looks does impact user behavior (which may or may not subsequently impact search performance), the look of your site is not that important to search engines. That’s not to say that making it sexy isn’t a good idea!
