Researchers Question Yahoo Answers
Stanford University researchers analyzed 10 months worth of content from Yahoo’s popular Answers service to learn about the people using it and where it might go in the future.
When we first covered Yahoo Answers in its December 2005 debut, we saw a striking similarity to Usenet. Instead of newsgroups, Yahoo provided Answers with web-based tools for asking questions, providing answers, and rating the responses questions received.
Yahoo Answers proved so successful that even Google could not compete with the service and shuttered their competitor at the end of 2006.
The appeal of Yahoo Answers drew the interest of researchers at Stanford. Resource Shelf picked up on their paper, Questioning Yahoo! Answers.
The researchers cited the rise of various forms of social media sites, like Flickr, Facebook, and StumbleUpon, and attempts by Microsoft and Amazon.com to mimic the success of Yahoo Answers with their services.
Among their findings, the researchers tell us some topics are more popular that others. From August 2005, the start of testing with Yahoo Answers (it did not open to the public until December 2005) through May 2006, Love & Romance proved the most popular topic, followed by Entertainment. The least popular topics were of a local nature: Dining Out, and Local Businesses.
When it came to participation, the same pattern seen in social media happened with Yahoo Answers. People were more interested in receiving than giving, according to the researchers:
Overall, we observe three interesting phenomena in user behavior. First, the majority of users are askers, a phenomenon partly explained by the role of the system: Yahoo! Answers is a place to “ask questions”.
Second, only a small fraction of these users gets involved in answering questions or providing feedback (votes). Hence, it seems that a large number of users are keen only on “receiving” from rather than “giving” to others. Third, there is a small portion of the user population that provides answers or votes.
They also noted the difference in the amount of participation, depending on the topic:
General, lighter, topics, which interest a large population, such as Love & Romance, attract many inquiries. On the other hand, topics that do not require expert opinions, such as Social Science, or that can be answered based on personal experience, such as Pregnancy & Parenting, attract many answers.
The researchers noted a lot of ‘noise’ in the current iteration of Yahoo Answers. They suggested what a more ideal question-answering system should do for its users:
Ideally, it should facilitate users finding answers to their information needs and experts providing answers. That means easily finding an answer that is already in the system as an answer to a similar question, and making it possible for users to ask (focused) questions and get (useful) answers, minimizing noise.
Their recommendation may be in practice already, at a site they did not mention in their paper. We have seen the model in action at Answerbag, which does try to route people to existing answers based on their questions.
The folks at Resource Shelf also noted another omission in the research paper. "We wonder if the researchers were aware of library virtual reference services and how they compare to what Yahoo Answers provides?" they asked.