Right at the beginning of the interview, BigSnoop president and founder Tamir Dinar clearly states, “We’re not going for the number of eyes model. We’re strictly infrastructure.”
”Globes”: Which field is BigSnoop’s technology aimed at?
Dinar: ”We tackle transient data. In the past, people complained about the lack of information on the Internet. At present, and in the future, the world is headed for a constant bombardment of enormous amounts of information. This information is vital to a very large range of sectors. The problem is that most of the relevant information cannot be found quickly, certainly not in real-time when the information is formed.
”We’re talking about information not found in static pages, such as web pages. BigSnoop enters the picture the moment dynamic information, which changes at very high speeds, is required. There are currently tremendous amounts of this sort of information on the Internet.”
What does BigSnoop actually do?
”We developed and are continuing to develop technologies for searching and retrieving transient data in real time. In other words, we enable constantly moving information to be found, for example information found in chat rooms, discussion groups and push announcement. The information processing takes place in real time, i.e. as the information comes along. It is at the processing point where the biggest technology barrier lies for anyone trying to deal with it.
”The processing includes real time indexing into categories and the transfer of the relevant information to the searcher. We place strong emphasis on relevance. In fact, BigSnoop’s goal is to organize the chaotic world of transient data.
”Information can be retrieved from the Internet and from closed systems, such as call centers. The moment certain information is wanted in real time, our system can provide it.”
How is it applied in commercial enterprises and what is the difference vis-a-vis the existing situation?
”Organizations and corporations can receive financial and other types of information “out there” on the network at that particular moment, using our system. There is currently no possibility of receiving information on a subject being discussed in all the different chat discussions and forums. Existing search engines provide static information, i.e. web pages. One of the major problems with the relevance of static information is the time span it takes for the update to be found by the search engine. At the moment, the information will be retrievable two weeks after the request, in the best of cases.”
How does it work?
”It’s a learning system and is therefore operated through NLP (Natural Language Processing) and pattern recognition techniques. As the information is being searched for, the system supervises and monitors the information requested. From the moment a query is submitted, it scans all the active information sources, not one by one, but simultaneously.
”As it scans, it reads and is capable of understanding the meaning of every word and the context, and can identify repeated behavior patterns on the part of the user and the information. If, for example, there’s an explosion of information on a certain matter, the system understands that a trend is emerging here and that something significant is going to take place.
“At the moment, since we are talking about enormous volumes of information, which under normal circumstances it would be impossible to gather, we developed a state-of-the-art approach to solving the problem. We can gather all this information for a certain period of time, both for answering the query and for clipping. For example, information about when and how a certain company was mentioned in the past month can be obtained, which could be helpful in decision-making.
What were the obstacles you were obliged to deal with?
”One of the biggest problems at the moment, making it very difficult to find information, is the incorrect phrasing of the query. Our technology overcame this. It manages to understand what the user really wants, without demanding the question be put in a certain way. The user can ask the question in ordinary language and the system knows how to track the most accurate answers for him, i.e. it does not retrieve irrelevant answers. This is possible, thanks to NLP. The system currently supports an extremely wide range of languages, including non-Latin ones.”
How does it work from the user’s viewpoint?
”We’re developing a flexible platform, providing the user with a series of screens through which he can submit his queries and receive the relevant information.”
Who are your potential customers?
”The system is designed for call centers, business information centers, large communications entities of all types, the financial market, including investment players and more.”
Who’s behind the development?
”We recruited the very best brains. We now have a 16-strong development team, all technology experts, including Prof. Alon Itai of the Haifa Technion, an NLP expert; Prof. Assaf Schuster of the Haifa Technion, a distributed-systems expert who set up and heads the Technion’s laboratory in this field; Dr. Daniel Keren of the Haifa University, a pattern recognition expert; and Dr. Yuval Saar, a data mining expert. Incidentally, all four work full-time in the company, not as external consultants.”
What is your potential market?
”I can’t speak about numbers, but in principal, we’re working in a world rapidly marching towards real-time applications. In the not-too-distant future, both open and closed systems will not look as they currently do. We can already today see that the more important and qualitative information is found in the constant flow on the Internet, not in static sites. This is the direction the market is taking, and it is already thirsting for systems that can provide possibilities for finding the relevant information at any given moment.”
BigSnoop has just started a first financing round, in which it hopes to raise $6-8 million. “The current round is for venture capital funds and a strategic partner. In addition, we’re in the middle of recruiting a general manager for the company,” Dinar says.
Dinar rejects the possibility that due to the subject’s sensitivity, forum owners like Yahoo! will prefer to be extremely cautious and nevertheless block the ability to scan dynamic content in its sites. He says that entities such as Yahoo! are interested in having the material they provide in their sites scanned.
|
Business Card
Name: BigSnoop
Founded: June 2000
Product: Transient information retrieval
Ownership: Argoquest and company employees
Competition: Potentially every player in the data search and retrieval sector, e.g. search engines.
web site: www.bigsnoop.com
|
Published by Israel's Business Arena on 20 December, 2000