IndustryBehind the Scenes in the Search Engine Labs

Behind the Scenes in the Search Engine Labs

Most of the major search engines have research units and live testing grounds for new products and features. What kinds of projects are they currently working on?

Most of the major search engines have research units and live testing grounds for new products and features. What kinds of projects are they currently working on?

A special report from the Search Engine Strategies conference, August 7-10, 2006, San Jose, CA.

Discussing new and forthcoming developments at the search engine laboratories were panelists James Colburn, Product Manager of adCenter, MSN Search; Bradley Horowitz, Vice President of Product Search at Yahoo; and Peter Norvig, Director of Research at Google.

Microsoft adCenter Labs

According to Colburn, Microsoft’s adCenter lab has four areas of focus:

  • Paid search
  • Contextual ads (including text mining and concept hierarchy)
  • Behavioral targeting (including age and gender prediction)
  • Emerging markets (for example, video ads)

What is great about the adCenter laboratory site is that it offers a wide variety of useful tools for keyword research aficionados.

For example, one tool available at the adLabs site is the context-based acronym resolution tool, which expands abbreviations based on related query words. For example, if you type in the abbreviation “ROI” (without the quotation marks) into this tool, you will find that the most common use for this abbreviation is “region of interest,” not “return on investment” as marketers might expect. Likewise, if you type in the abbreviation for the state of California, “CA” (without the quotes), you will find that this abbreviation is most commonly used for the word “cable”.

Another tool that is useful for keyword research is the keyword mutation detection tool, which shows common misspellings of keywords. Just for fun, I decided to type in the word “Google” (without the quotes) to see the company name’s common misspellings and was quite surprised to see 26 common misspellings. If you are doing search marketing for an industry whose target audience frequently misspells important keywords, such as the health/medical industry, this is a very useful tool.

The keyword group detection tool will also detect common misspellings, but more importantly, it will also detect other word types that searchers use. “For example, you target audience might use the word ‘Halloween’ as a keyword,” said Colburn, “but our tool might detect a holiday theme.”

Yahoo Labs

According to Horowitz, Yahoo labs’ four areas of focus are community, microeconomics, information navigation and search, and user experience. For example, when analyzing communities, Yahoo researchers look at the following information:

  • How to know what to believe
  • Trust models online and propagation
  • What makes communities thrive and wither
  • Tagging images and videos, and sharing these file types

Horowitz said that Yahoo believes in better search through people. “Writing algorithms to understand what is in an image is extremely difficult, for example,” he said. “People plus algorithms is better than algorithms alone.”

Algorithms have evolved considerably since 1995. Horowitz characterized search algorithm evolution into four phases:

Phase 1: Human editorial (Yahoo directory). “In 1995, this worked because the sources were trustworthy and authoritative,” he said. “The problem was that it was not scalable.”

Phase 2: Mass automation (AltaVista). With mass automation, search engine software engineers were able to automate process of acquiring information on Web pages. “This process was better because it was scalable, simpler, and more granular,” he said, ” but the information was not authoritative and subject to many types of search engine spam.”

Phase 3: Topical analysis (Google). Topical analysis uses link structure to confer authority. However, topical analysis is still subject to other types of search engine spam, such as free-for-all link farms, which can confer authority on poor-quality Web pages.

Phase 4: Social search. Builds on all earlier technologies. “Subjective queries rely on domain expertise,” said Horowitz. “With social search, we are handing back domain expertise to the users.”

One instance of social search in action is Yahoo Answers, which connects a searcher with a question to a community of people best suited to answer a question. For example, if you accidentally drip hot wax from a candle onto your floor, you can use Yahoo Answers to find out how to clean it up.

Google Labs

Google Labs has been a playground for search engine optimizers and marketers for a long time. Google Labs offer a wide variety of searcher toys that are not quite ready to be unveiled on the main Google site, and some developments eventually become a significant part of the Google landscape. “For example,” said Norvig, “AdSense came out of Gmail. We looked at matching ads against words versus matching ads against the page content.”

According to Norvig, some forthcoming items from Google Labs include:

Image processing. Google is now capable of face localization. “There are important pixels in identifying a face,” said Norvig. “Currently, we can identify a male or female face, but we cannot yet identify a specific person.”

Fact extraction. There are well over one million facts available, such as a city or country population. “If you identify writing patterns ahead of time, go to the Web, find examples, and cluster them,” he said, “you can more easily present this information to searchers.”

Statistical machine translation. “Using Google language tools, a model of the English language (greater than a trillion words) and a model of translation (such as a news article in both English and Arabic),” he said, “translations can become more accurate.”

Whether or not the developments from search engine laboratories are actually incorporated into search engine algorithms or the main site, becoming a tester or participant in these laboratories provides useful information to the search engines. Based on the feedback given from the interfaces and tools, search engine laboratories build tools that provide a better search experience for all users.

Shari Thurow is the Marketing Director at Grantastic Designs, Inc. and the author of the book Search Engine Visibility.

Search Headlines

NOTE: Article links often change. In case of a bad link, use the publication’s search facility, which most have, and search for the headline.

From The SEW Blog…

Headlines & News From Elsewhere


The 2023 B2B Superpowers Index
whitepaper | Analytics

The 2023 B2B Superpowers Index

Data Analytics in Marketing
whitepaper | Analytics

Data Analytics in Marketing

The Third-Party Data Deprecation Playbook
whitepaper | Digital Marketing

The Third-Party Data Deprecation Playbook

Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study
whitepaper | Digital Marketing

Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study