As for the Mathematical Model, it's between the algebraic and probabilistic model. As for the properties, it' in the last column : transcendent term-independencies.
algebraic because of the fact that we use a scalar to rank the documents?
probabilistic because of the probability ranking principle: PRP --> yes sure.
Instead of algebraic, I would say that it's informational since we use also tf-idf as a basis, which is a way to assess the relevance of a given document against a given term, from an information theory stand point; in fact, it's related to information entropy, right?
So in short, it s an hybrid between probabilistic and informational, probabilistic-informational, I would say.
As for transcendent term-independencies, what do you mean?
The ascot project is the first project started by bunkersofa llc. It's started at the bunker (ascot meguro) and aims at getting rid of the search bar. The search bar has been used since information retrieval became a real science in the early 70's. It is very suprising that almost no innovation has been done in this regard for 3 decades. The ascot project is therefore an project that is trying to do something about it and hopefully we'll be able to give a better search experience to the whole humanity soon
contributors
Pascal Hamonic, researcher, entrepreneur, co-founder at bunkersofa llc and more importantly singularitarian: linkedin, twitter
Julien Hamonic, researcher, entrepreneur, co-founder at bunkersofa llc and more importantly singularitarian: linkedin, twitter
where would you place ascot system?
ReplyDeleteI think we are more related to semantic web than search or both.
As for the Mathematical Model, it's between the algebraic and probabilistic model.
ReplyDeleteAs for the properties, it' in the last column : transcendent term-independencies.
algebraic because of the fact that we use a scalar to rank the documents?
ReplyDeleteprobabilistic because of the probability ranking principle: PRP --> yes sure.
Instead of algebraic, I would say that it's informational since we use also tf-idf as a basis, which is a way to assess the relevance of a given document against a given term, from an information theory stand point; in fact, it's related to information entropy, right?
So in short, it s an hybrid between probabilistic and informational, probabilistic-informational, I would say.
As for transcendent term-independencies, what do you mean?