About ZeBAze Computing

ZeBAze Computing is a young and dynamic start-up specialized in artificial intelligence and datamining technologies. Our team is dedicated to offer you a constantly evolving solution with a simple philosophy : an extremely easy-to-use software doing extremely efficient queries of your databases. The datamining possibilities offered by artificial intelligence are endless and we will constantly adapt them to add more features and functionalities to ZeBAze.

ZeBAze Computing was founded by Matt Cazaux, an artificial intelligence specialist who always had in mind that the strict rigidity of database querying was a major problem to solve. He officially started his research & development activities in 2004 to adapt the flexibility of artificial intelligence techniques into specifically designed datamining functions, finally leading to the release of ZeBAze.



    About the Artificial Intelligence technology used by ZeBAze

The most important and exclusive feature of ZeBAze is its use of artificial intelligence techniques and datamining functions to query your database and rank the search results according to your preferences.

All the interfaces of ZeBAze have been designed to be extremely easy to use and you do not have to worry about scientific concepts. You just have to define your search preferences and click a button to start the search engine.

However if you want to learn a little more about the concepts used by ZeBAze, this chapter gives you a "visual" understanding of the different steps followed by ZeBAze to rank your search results :

The main property of number and date values is that they are ordered, which means that they can be placed on a graduated line and compared between them.

For any number or date field, ZeBAze allows you to define what is the value that you want to target. This best value is used as a reference to rank the search results.

For each database row, ZeBAze calculates the distance (referred to as D) between the current value of the field (referred to as X) and the best value of the field (referred to as B).

    

In the following example, we define a best value for five fields (F1 to F5) of the database.

For each row of the database, a specific set of five distances (D1 to D5) represents the difference between the current values of the row (X1 to X5) and the best values (B1 to B5).

These five distances are used to determine the proximity of each database row with your search preferences.

    

The first step is to create a single structure enclosing the five individual distances as a representative group.

The difficulty is that the structure must maintain a proper correlation between the five distances, depending on their individual sizes, on a one-to-one and one-to-all basis.

ZeBAze uses a merging function that creates a convergence structure in which each distance's representation is influenced by the other distances, and can be moderated by scale values.

The key role of the merging structure is to work on the five distances as a whole and not as semi-individuals.

    

ZeBAze allows you to assign a specific importance to any of the fields with the following slider :


The difficulty is that these importance values have to influence the calculation of proximity for each database row without losing the overall representation of the merging structure.

ZeBAze uses a ponderation function that integrates the respective importance value of each distance inside the merging structure.

    

The final step is to evaluate in which proportion the current database row is similar to the search preferences.

ZeBAze uses a reduction function which decomposes the merging structure and analyzes its entire composition depending on its properties.

An overall proximity value (decimal number) is finally calculated and determines the degree of similarity of the database row with the search preferences.

    

This process is repeated and a proximity value is assigned to all of the rows in your database. ZeBAze is then able to sort the rows based on these values and finally rank your entire database according to your preferences.

The artificial intelligence techniques and algorithms used by ZeBAze are in fact slightly different and more complex than the ones presented in this example, but the principle is the same : the ability to flexibly process several search preferences as a "whole" solution, even if all these preferences are not equally important.