By Dongfei Yu

· Introduction

• • • • • • • Due to my interests and previous research focus on web search and mining, so it’s easy to connect my proposal with web search. Generally speaking, what I want to do is making current query suggestion on commercial search engine more natural. We will do this with two main steps:

            • 1. Given a query, we will do query clustering from query log and select top k based on semantic similarity. There are many successful algorithms to achieve goal.

            • 2. Then we designed a friendly UI such as “3D rotating tag cloud” to display our suggestion depend on user’s current query.

We give a simple instance for describe my minds.

When user given a query such as “google”, we can get some related queries such as “google map”, ”google images”, ”google fondos” and so on. Besides, we get a similarity measure on each related query. Then we can create a 3D query cloud to show related recommendation. Different size of query represents different similarity with original query.

· NABC analysis

– Need

We locate our product as a feature on commercial search engine. As we all know, query suggestion is common and useful when we search something. On one hand, query suggestion will reduce the typing time and speed up what we want to find. On the other hand, query suggestion will give assistance while the user can’t exactly express their demand with a keyword or the return result doesn’t satisfy the user. In fact, exploratory search is a very hot research fields.

– Approach

There are many related research and method on query suggestion. Most commercial search engines support such features. When user type something, some related queries will appear in Droplist. So user can select instead of typing what they want to search.

But we want to do more and offer a more cool and natural approach to accomplish query suggestion. “3D tag cloud” seems to be a possible way. We will create a “3D query cloud” on search engine, so user will easily find more exact query from the cloud when the original query can’t meet their demand.

– Benefit

The major feature on our project is cool but useful. It will give a friendly user experience while searching. And it won’t be very complex.

– Competitors

To our best knowledge, there are no such features on the major search engine, such Bing/Google/Yahoo. But query suggestion is common and successful on it. What’s more, 3D tag cloud tech is not very difficult, and has been used in Blog tag/product recommender system. So we combine the query suggestion and “3D query cloud” UI, it will be a creative feature.

· Conclusion and General schema

In a word, we want to put query suggestion into 3D query cloud. So our work will focus on two sides: a well-performed query suggestion algorithm and design 3D query cloud interface. We will have an alpha version demo within one month, and testing will occupy another month. More idea and improvement may be proposed later.