The Anatomy of a Large-Scale Social Search Engine
This article introduces a new social search engine named Aardvark. To tell the reader how this
engine works, the author explains it in details: how to establish the model, how to analyze
question, how to find a right person, how to make a ranking algorithm and how the user interface
works. Finally there is an evaluation for this social search engine and analyze its highlight merits.
One interesting thing I found is the idea of this social search engine. When comes to search,
people always focus on the finding result: Does it satisfy what they want to know. However,
seldom people will think about how to find the result, based on what? Documents or a person?
This engine provides novel idea that it lies in finding a right person. It just like when you have
trouble, lost or in danger, you may only want to ask help from police. Furthermore, its user
interface and utilize experience seems like using an instant messaging tools. You can use it via
email, mobile, Google talk, MSN, or its website. As it is said, almost 90% asking can get response
in 5 minutes, there are a broad range of answerers, people are indexable. Since Aardvark has
been acquired by Google from 2006, I believe the number of their customers should be inceased.
However, there is one thing I doubt on this social search engine. It is said that there are too many
documents to index, and once the answer is indexed, it may not be the correct one. But how do
you make certain that the person you find is the right person to answer. People tag themselves
do not mean they are expert on that area. Ok, you may say if the people don’t know the answer,
he/she can choose not to respond. But how to make sure the person who answers it finally is the
In my view, this social search engine has some limitations. Firstly, it is limited in a particular circle.
Once you ask a topic, question will be classified in a certain area, and the system will find the
answerer from this circle. We cannot find more suitable answers if the searching range is
restricted. Secondly, it is too conversational. I would like to prefer the sharing information
environment that people can find out what they want to know, thus they don’t need to ask the
same question again and again. Peer to peer increases system workload and may decrease the
effective and efficiency.
To improve this social search engine, I would like to suggest add search bar onto the interface so
that people can search the topic they want to know instead of ask the particular person.
Additionally, I wish this engine can support answering anonymously and add video function when
answering the question.