[ About Us | Popular | Marcom | AdNet | IChannel | Glossary ]

May 19, 2008

Live: Google Health Launches At Factory Tour of Search

Mark Hendrickson

Google has invited the press to its headquarters in Mountain View today for some presentations on a variety of search-related projects...



The event is about to start. There are representatives here from Cleveland Clinic, Harvard Medical School, MedCo, Quest, Walgreens, CVS, Longs, RxAmerica and more. I hear that Google Health will actually launch today in a couple hours. This comes after expectations of it launching this past January(and in February, and in May 2006 for that matter). Update: The Google Health sign in screen is live here already.


9:50am: Event has finally begun. They’re beginning with a general overview of Google’s search properties. R.J Pittman on stage talking about the breadth of search, and three properties in particular, starting with Image Search.

Over 300 million digital photos taken every day, 100 billion per year, over half a trillion images in circulation by 2009. Wants to see Google have over 1 trillion images to search from one day.

9:55am: More users turning to image search than ever before. Today hundreds of millions of image searches per day. Important to help users navigate, so innovation is hugely important.

Likeness of an image is a difficult thing to figure out but possible with Google’s image similarity technology today. Google also has some of the first and best tech for object recognition. Adding this tech will be increasingly important with more photos online. People also like to search for people, so facial recognition important as well. Working to make tech for facial recognition one click away, so you can distinguish between, say, Delorean the inventor and the car.

9:57am: Gigapan project meant to handle big image sizes. Also working on geo-tagging and location-based imagery.

People using image search for more than just looking for pretty pictures. It’s being used for recipe hunting, travel planning, diagnosing health, shopping, etc. Relevance of commerce-related activity to imagery is increasing.

10:00am: Google has experimented with ads around images but have had to struggle with how to do it right, since it’s not easy. Announcing today a new suite of image related experiments. Pairing images with images — display ads with image search.

Examples show sponsored image results with AdSense-like descriptions, displayed along the top and on the right. Doesn’t appear to be launching publicly anytime soon, Google will “keep us posted” on how these experiments go.

10:03am: Now on to Google News. The “story cluster” is very important to the organization and display of info.

Local news is a hard problem to solve. Goes beyond zip code or city name. The way we are approaching this is quite novel; leveraging power of clusters to aggregate articles and reaffirm approximations.

10:05am: Quote search is possible on Google News to find quotes from particular people in news articles. Working to leverage clusters for quote search as well to improve its accuracy.

10:06am: Google Finance launched about a year and a half ago, simple and powerful stock market tool. Led that introduction with powerful visualization tool; important because goal is to help all users navigate complexities of market.


Mayer: Google Maps a “big home run” for Google. Inviting Carter Masian on stage to talk about Maps.

Masian talks about “building a map that contains the world’s information”; Covering three things: why that’s hard, the state-of-the-art, and the newest innovative work.

10:17am: Why it’s hard: Masian discusses the issues surrounding the identification of places, including international complexities. Not always clear what people are trying to find (”New York NY” that city or the hotel in Las Vegas, for example).

There’s a need to collect local information from local sources — the long tail of geographic content. Need ways to gather information from locals. Go to maps.google.com/recentedits to see additions being made to Google Maps in real-time.

10:23am: Talking about how you can use Street View to see what turn-by-turn directions actually look like. Also discussing coordinated efforts to map and annotate all features globally (reminds me of Earthmine). We “have the base canvas” - now how do we enable people to “annotate that canvas”. Annotations include tourist destinations, transit maps, etc.

Over 350m unique user activations, over 50m unique maps/earth users per month. Annotation has exploded over past year alone (shows map that highlights areas that have been annotated).

10:29am: Continuing discussion of the different ways Google Maps has been annotated for particular uses. Shows how images can be integrated into maps from sources like Panoramio.

10:31am: There are lots of hard queries, some we handle well (”beachfront hotels los angeles”, “soccer field burlingame”) but others need work (”marin headlands hiking”, “kansas state”). Sometimes a little too literal with search queries, as with “kansas state”.

10:34am: Even as we tackle these query problems, there will be a whole other set of problems when street view comes into play. Such as, how do you search farmers’ markets, parking restrictions, opening hours for stores, etc.

10:36: Masian wraps up his talk, Mayer introduces presentation on universal web search by Johanna Wright.

Wright starts off by talking about how Google guesses what users actually mean, even when they spell things wrong or don’t know exactly how to structure their search.

Universal Search - effort to bring all types of search results onto same page (whether webpages, pictures, videos, news, books, etc). In 2007, launched universal search for videos, maps, news, books, and images. Have added products and blogs since then.

Three challenges to universal search: infrastucture, ranking, and user interface.

10:40am: Universal search takes much more computational power than simple web search (has to retrieve from multiple sources). Questions about how you compare books to blog posts, images to videos, etc. Have been working to launch universal search worldwide, available now to users around the globe.

10:42am: This year added review content to local search results, blog results to universal search, and improvements to video results. Crawler has also been improved to surface video hosted on non-Google sites better.

Biggest improvements have been made in image and video results in universal search; now showing more results for lesser-known photographers and video producers.

10:47am: Trystan Upstill invited on stage to talk about the internationalization of search.

If you search for “tax” in the US, the results should be much different than when you search for “tax” in the UK or Australia. Different resources show up to help people deal with taxes.

Country localization vs. language localization. Concerned not only about readability across languages but watchability as well, for when it comes to video. Want to show English videos about Enrique Iglesias, for example, even in a Russian search.

10:53am: Cross-language search results help non-English speakers find good information that’s only available in English.

11:00am: Pandu Nayak on stage talking about various ways in which Google needs to determine what users actually want to find with their queries. Lots of examples of how queries can be confusing.

11:22am: After a brief break, it’s time for Google Health. Mayer back on stage, recapping the presentations thus far.

11:28am: “Future of Search” - we imagine search in future will be experienced through many modes (cell phones, cars, etc). More modes means more complexity (search when people are moving has different dimensions).

Element of personalization - know a little more about you, build that data into the search engine...

Ed: Google keeps eyes on the knitting.

Google Health Launches, But Gets A Second Opinion On Terms Of Service

A lot has changed since the concept of Google Health was first broachedin July 2006. After months of preparation, the site was finally launched on Monday. Google's (NSDQ: GOOG) pitch to users is that they will be able to keep all their health information—including details about users' doctors, hospitals, medical history, lab tests, etc…—while being able to search for additional resources related to your particular medical profile.

-- Terms of service: As the terms state, "When you provide your information through Google Health, you give Google a license to use and distribute it in connection with Google Health and other Google services." Google is not a "covered entity" under the Health Insurance Portability and Accountability Act of 1996 and therefore, the provisions of that law do not apply to what Google does with your info. That means if you disclose details such as whether you have AIDS, HIV or any sexually transmitted disease, have been treated for drug an alcohol abuse, have had an abortion, or have a genetic predisposition to any diseases, you have to just trust Google to keep that information secure. A webcast with the heralding Google Health is available here.

No comments:

Post a Comment

Comments accepted immediately, but moderated.

Support Our Sponsors: