* Race is on to catch up with Amazon.com
* Wal-Mart builds search engine from scratch in about 9 mos
* EBay replacing decade-old engine with Cassini by 2013
* Google uses visual-recognition tech for shopping search
* Search expert Etzioni works on shopping search site Decide.com
By Alistair Barr
SAN FRANCISCO, May 4 (Reuters) - EBay Inc and Wal-Mart Stores Inc are developing new Web search engines to better compete against Amazon.com Inc in the fast-growing e-commerce market.
As more people shop online, they often end up at the top of a website typing in a product name. If they cannot find what they want quickly, they will likely go to a rival website or venture into a physical store.
“Amazon is on version 8.0 of search,” said Scot Wingo, chief executive of ChannelAdvisor, which helps merchants sell more online. “EBay is at 2.0, but they are thinking about how they make this huge leap to 3.0.”
The stakes are high because e-commerce is a huge, fast-growing market, putting billions of dollars in sales up for grabs. U.S. retail spending online grew 13 percent to $161.5 billion last year, according to comScore. Physical retail sales are much larger, but the sector is struggling to grow and losing share to online operators.
EBay’s search technology, known as Voyager, dates back to the first dot-com boom a decade ago. After the company appointed Mark Carges as chief technology officer in 2008, he tested eBay’s search engine by typing in “iPod.” A car topped the list of results because the seller noted in the listing title that it came with an iPod adapter.
“Search was clearly broken in 2008,” Carges said.
Since then, eBay has gone on a hiring spree to fix search. The number of employees working in that area has tripled to more than 150. EBay also poached several engineers from Microsoft Corp’s Bing search unit, including Ken Moss, who runs the Seattle office, and Hugh Williams, who oversees eBay’s new search engine, Cassini, to be rolled out in 2013.
“More customers, plus better search, means people buy more stuff,” eBay Chief Executive John Donahoe said in a recent interview.
Cassini will trawl full product descriptions, rather than just the titles of listings, and match search queries to photographs of products, while taking into account information about the seller and the buyer.
By crunching data on what shoppers have bought and browsed on eBay in the past, Cassini search results should be more tailored to their intent. For instance, if a shopper types in “HP,” Cassini will know if the person means horsepower or Hewlett-Packard Co, Williams said.
“Voyager is pretty literal. It takes a query and matches it faithfully against the title of items. It’s not intuitive,” he said. “Cassini will take the user’s query and understand that.”
The search engine project takes time because eBay’s online marketplace has so much variable information from millions of listings that are described differently by each seller - something known as unstructured data in the tech world.
In contrast, Amazon typically starts with a catalog of items it has for sale, including strict product descriptions, which are easier to search.
Wal-Mart recently launched a new search engine on its website that was built in less than nine months with 10 to 15 developers, according to Anand Rajaraman, who helps run the discount retailer’s Silicon Valley tech arm @WalmartLabs.
The new search technology focuses on groups of related terms and phrases people use when describing products, rather than matching queries to exact words in listings.
“Wal-Mart’s search knows that a backyard chair is the same as a patio chair or a garden chair,” Rajaraman said. “These product listings will come up on that search too.”
If eBay and Wal-Mart can vastly improve search on their websites, that could eventually threaten Google Inc, the world’s leading Internet search company.
“Google doesn’t want you to go directly to eBay to search for products,” said Oren Etzioni, a search expert at the University of Washington’s computer science department. “A lot of what funds these search efforts are e-commerce ads. If eBay and others stop advertising as much on Google, that would be a problem.”
A Google spokesman said the company is “very focused on product search.” One of Google’s latest upgrades to its shopping search uses visual-recognition technology to track down products, especially in the apparel and fashion categories.
Type “red dress” into Google’s Shopping page and frocks sold by retailers like J.C. Penney Co Inc, Nordstrom Inc and Neiman Marcus show up. Shoppers can refine the results by choosing silhouettes, such as “Empire Waist” and “V-neck” on the left side of the page. Google matches those phrases with photos of red dresses with those shapes.
“The stakes have never been higher for whoever can deliver a high-quality shopping search experience for consumers,” said Etzioni, who has advised firms including Google and Microsoft.
Etzioni founded airfare search company Farecast, which Microsoft bought in 2008. His latest project is Decide.com, a product search website that trawls thousands of news articles and blogs to advise shoppers when prices may change or new versions of gadgets may come out.
“I‘m putting my money where my mouth is. We’re building a next-generation shopping search ourselves,” Etzioni said.