Ebay facilitates online trading via large data

On the Internet, data is money. For this, commodity http://www.aliyun.com/zixun/aggregation/34926.html "> trading website should be most experienced." As the world's largest auction site, Ebay has a staggering amount of data. So how do companies use this data to promote online trading? Recently, Ebay's vice president for search and platform, Hugh Williams, shared the company's approach at the Big Data conference in Sydney. Summit.

The Ebay site has 180 million active users. At any given moment, about 350 million of the merchandise will be sold. Ebay's Auction search engine produces 250 million searches a day. Hugh Williams says the company's raw data on the Hadoop cluster and Teradata servers are petabytes. Through the use of these data, companies can increase online trading volume, to guide users to buy more goods.

Hugh Williams mainly talked about search engine optimization. Through the analysis of the data, the company can grasp the user's behavior patterns, so as to optimize the search results. "If you go backwards a few years and you use an Ebay search engine, you'll find that it only understands the literal meaning," Williams says. "You can enter text in a search engine and it will look for it literally, but it doesn't really understand the deep meaning of your thoughts."

Now, Ebay wants to make search engines more "intuitive". As an example, Ebay found that those who want to buy Pilzlampe, if they enter the "Pilz Lampe", are more likely to buy goods, because the search engine will give more results, so the search engine by adding a space in the word, can increase online transaction volume.

Ebay will change or rewrite the user's search requests, add synonyms or replace statements, and give more relevant results based on the information it has. In addition, Ebay also uses big data to predict whether a product will be sold, how much it sells, and then adjust the product's ranking on the search engine.

Adding factors to change search results is risky, Williams says. Companies usually need a few months to add a factor, and they don't know if it really works for users. To do this, Ebay tests a small number of users and then observes their reactions.

In the process of improving search engine, it is a challenge to judge the situation of user search request. For example, when users search for "Geelong Cat team" (Geelong Cats), search engines may make "cat" as a keyword, and then search the category of pets, which is useless information for users. "There are a lot of tiny problems that can happen on a scale like ours, so people like data specialists need to look at these things," he said. Williams said.

(Responsible editor: Schpeppen)