Semantics to enhance BSS/OSS
Javier Martínez Elicegui
Of course, day after day the number of applications of semantics in the enterprise is continually growthing. This post shows a case of application of these technologies on BSS/OSS systems.
BSS/OSS systems are usually very complex systems, with lot of interfaces, lot of different users, lot of applications related… Besides this, in the last years there has been a constant competitive preusure on these systems requiring them more functions and less cost to be managed. This fact has marked a quick evolution in this kind of systems:
- Software that needs to be modified each time a new requirement is requested.
- Software including a configuration file holding the parameters to adjust to the different needs of each installation or context.
- Software incorporating configuration management functionality.
- Software with great complexity and high parameterization capabilities. In this case the incorporation of new scenarios/processes does not need a new version of the software, but it demands a complex configuration task (e.g. for each new product there are setting with all features of sale, provisioning, billing, risk control, bundling with other products, etc.).
- Software using a Knowledge Database that provides flexibility to pick up all kinds of concepts, relationships and patterns that the administrator needs to use, in a consistent and not redundant way, as seen in the next figure. This requires the introduction of a semantic layer over the relational database.
Telefónica has reached this last stage in the evolution of BSS/OSS using Semantic and Data Mining techniques in its tariff system. The tariff system is complex to configure, hence it requires very specialized people, errors are common, and they are expensive to maintain.
A first step to implement the semantic layer in a system already running is to discover patterns in the information already there. To achieve this, we have used Data Mining methods to carry out an in-depth analysis to discover association rules, classifications, clustering, … Once validated by users, such rules are stored onto the Knowledge Database, transforming implicit knowledge, perhaps already forgotten by the users, in explicit knowledge modelled to ensure that it will not be left out by mistake from that moment on. The administrator can incorporate, new business rules that enable the system at any moment.
- Explicit Knowledge obtained: the tariff logic is now explicit, easily verifiable and editable by administrators, unlike tariff tables where the values of the coefficients lie but not the logic applied to obtain its values.
- Easier to maintain: Business Rules are simple atomic reasoning that combine to calculate coefficients. A simple change in a business rule may affect hundreds or thousands of records in the tariff tables.
- Risk Control: The knowledge database detects inconsistencies between rules, which prevent many of the current errors. In addition, the knowledge database contains few hundreds of business rules that can be easily verified by a person, unlike tariff tables where there are many thousands of cases that escape effective control.
- Reduction of time to market: the time required to update the system when a new service is introduced, is reduced extraordinarily.
Leave a Reply
Follow comments:
