Powerset vs. Cognition: A Semantic Search Shoot-out - GigaOM
The comparo part of the above is a good read, though given that both Powerset and Cognition must be improving stuff behind the scenes, and big G has so much "legacy" to deal with, I'm not sure there are any conclusions to be drawn.
Whats more interesting is the second part of the article where Nitin discusses what kinds of disruption search companies could shoot for. At Ziva, we're doing a couple of things already, and then some!
a) Mobile specific focus. There's a lot of folks out there who have not used "web search". And I include a fair number of techies in this - given how many questions I see in my mailbox, IM windows and even when people call up to ask for stuff thats actually possible to find on a search engine. There are those who "can search" and those who cannot, for a variety of reasons ranging from access to discomfort.
b) Answers, not links. Semantic Search/disambiguation/NLP etc are tech-friendly ideas that the end user doesn't really connect to. What they need is an answer to a need. On the mobile, its even more extreme - cause the effort a user usually needs to put into a web search engine to clarify sense etc is just not feasible on the phone. Also, not every question has an answer in a repository. To an end user, these are NOT separate use cases.
c) Instant answers, or surely-relevant one ? Like Nitin points out, we've too believed that not all questions need a resopnse in a flash. There are a huge range of response times which are acceptable for different kinds of questions. Involving the community expands the range of questions that can be answered to near-infinite, and the fact that the asnwers may come a few minutes later is more than compensated for by the relevance of the answer.
The comparo part of the above is a good read, though given that both Powerset and Cognition must be improving stuff behind the scenes, and big G has so much "legacy" to deal with, I'm not sure there are any conclusions to be drawn.
Whats more interesting is the second part of the article where Nitin discusses what kinds of disruption search companies could shoot for. At Ziva, we're doing a couple of things already, and then some!
a) Mobile specific focus. There's a lot of folks out there who have not used "web search". And I include a fair number of techies in this - given how many questions I see in my mailbox, IM windows and even when people call up to ask for stuff thats actually possible to find on a search engine. There are those who "can search" and those who cannot, for a variety of reasons ranging from access to discomfort.
b) Answers, not links. Semantic Search/disambiguation/NLP etc are tech-friendly ideas that the end user doesn't really connect to. What they need is an answer to a need. On the mobile, its even more extreme - cause the effort a user usually needs to put into a web search engine to clarify sense etc is just not feasible on the phone. Also, not every question has an answer in a repository. To an end user, these are NOT separate use cases.
c) Instant answers, or surely-relevant one ? Like Nitin points out, we've too believed that not all questions need a resopnse in a flash. There are a huge range of response times which are acceptable for different kinds of questions. Involving the community expands the range of questions that can be answered to near-infinite, and the fact that the asnwers may come a few minutes later is more than compensated for by the relevance of the answer.
.jpg)
No comments:
Post a Comment