FAQ

Here you will find answers to all the most frequently asked questions about Sight'Up technologies.


1. What does Sight’Up do?

Basically, Sight’Up develops infrastructure technologies which automate processes and organize large quantities of not structured information into suitable contents for search engines.
Sight’Up’s artificial intelligence engines can almost be regarded as operating software with additional intelligent functions. The core technology (IS) provides a platform for automatic categorization and the profiling of non structured information, allowing the automatic delivery of large volumes of personalized information. Our technologies are especially used by e-commerce companies and for incoming e-mails management.


2. What Sight’Up software is not.

They are not search engines but rather infrastructure tools which organize heterogeneous data for direct use by  search engines.
They were especially designed for the world of the e-commerce (product, news …) and for the management of  e-mail, both of which have the particularity of having practically non-existent grammatical structures. On the other hand, they are less appropriate for documentary structuring involving long and literary texts (long documentation, patents, books …).


3. What are Sight’Up’s markets ?
 
The main markets are those of e-commerce and the technology is especially suited to short non-literary texts. The engines are designed to give a high "degree of precision" while at the same time minimizing "noise" which is very penalizing in these markets.


4. How does "machine-learning" work ?

"Machine-learning" is a mechanism by which knowledge is acquired through experience. There are several learning models ; those based on statistics, logic, neuronal structures, information theory and heuristic search algorithms. Sight’Up has developed algorithms, based on the genetic behaviour of learning documents, which identify characteristics in the data observed in order to predict the behaviour of new data not yet encountered.

This technique is used for the following Sight’Up products:

Sightis : Categorization Engine
Taggis : Characteristic extraction Engine
MailRelation : Management of incoming e-mails


5. What are the differences between Sight’Up technology and rule-based / statistical approaches ?

• The rule-based approach, classifies documents according to on the existence or the absence of predetermined key words. This method suffers from several disadvantages, mainly that of the rigidity of the rules. Consequently, it is very difficult to create rules because they are not sensitive to the context of each document. The addition of a new rule for a counterexample can jeopardize the whole system. It should also be noted that the people in charge of the maintenance of the system have to possess both  linguistic and data-processing skills.

• The statistical approach uses the presence or absence of words in the documents to make predictions. This approach is only meaningful when applied to large populations, otherwise it is impossible to obtain  convergence. Thus learning involves processing a large number of documents and leads to a lot of human intervention. The majority of these types of approach do not take account of sentence structure such as for example the position of words in a sentence and this seriously impairs results.

Click on the link to learn more about benefits compared to other approaches .


6. Which languages are managed?

Sight’Up technology is completely independent of the language used. To start in a particular language, it is enough to learn in this same language.
This is valid for Romance languages, but also for Asian languages such as for example Chinese (traditional and simplified) Korean, Japanese and Vietnamese.


7. What does the system need to start working ?

Unlike other types of solution, Taggis and Sightis require only a minimal installation before they start working. This software can give a very good result with a small learning corpus ; 10 to 12 documents are quite enough. The small number of learning documents significantly decreases the effort required for the system to learn.


8. How do I integrate these engines into my existing applications?

Sight’Up engines can be regarded as technological bricks easy to integrate into the process of your trade. They include an API XML which makes integration simple with other systems. Management systems become "customers" of the engines.


9. Which APIs are available?

C, C++, Java and NET.


10. Which operating systems and databases are usable?

Linux, Unix and Windows NT/2000/XP.
The Sight’Up artificial intelligence engines function on all types of database which work in XML format.


11. How fast is processing?

Our artificial intelligence engines are developed entirely in C, which allows great portability and a very large processing capacity :

• Between one and two million documents per hour

 

Today Sight’Up manages daily more than 100 million documents in ten languages.


If you have any other questions : contact us