Back in March FICO released 6.7 of their BRMS Blaze Advisor. I got a chance to catch up with the release just recently. The focus of this release was the business user experience – an improved out of the box experience for business users. The business user experience has become increasingly critical as business users have taken more responsibilities in many business rules projects. In particular many business users are now managing lots of rules and doing this across multiple projects. This creates challenges such as helping users navigate thousands of rules in a repository, find related rules in these circumstances and manage changes that they have to make.
Helping business user navigate complex content involves two main areas. Filtering is the first of these. Filters have been available in Blaze Advisor’s development tools for a while and support sophisticated query-like structures to select a subset of a rule project or repository. These are now being applied to the rule maintenance applications as well. They can be defined by technical users and built into the web-based interface or can be defined on the fly by the users of the web based tools. These queries allow, for instance, tags or labels assigned to rules to be used as the basis for a filter (think Gmail labels). Filtering can be used to provide multiple views across the same set of projects or to quickly find an area that needs editing.
The second area of navigational improvement is designed to make it easier for non-technical users to manage rules that are being reused. When a rule set is reused in a rule project (a decision table that is embedded in another decision table for instance) the rule maintenance environment now displays a link. The rules can be previewed using the link from the place where they were reused. The navigation is direct – no need to understand the repository structure or project hierarchies. This can be important as the structure of a rule repository is driven to enable management of all the rules. A specific instance of reuse might cut across this and so navigation needs to be direct. As well as preview the business user can also decide to edit and save rules and then return (assuming they have the right permissions).
The next area of change is designed to help business user understand their changes. When business users make changes, especially to graphic metaphors like decision tables and decision trees, it is really important that they can easily see what changes are made. Traditional programmer’s “diff” tools tend to be confusing when applied to graphical metaphors as they compare the underlying or implied rules not the metaphors themselves. Even when comparing two rule sets a traditional diff tool is not that helpful. The new release has a nice color-coded graphical comparison suitable for all users. The cells, nodes, links or rule fragments that have been added, deleted or changed are highlighted right in the format used to write the rules – no XML, no geeky code-like descriptions. This is a very nice feature that will make it much easier for business users to see what changes they are, for instance, about to apply or what the cumulative set of changes made over the past month look like. This in turn will increase their confidence in their change and thus their willingness to take ownership of the rules.
There were also a few metaphor updates around readability. Decision tables now have frozen headers so they stick as you scroll and filters like the auto-filter in Excel. This allows you to select one of the values in a column and see all the rows that contain that value. Nice feature for navigating large tables. In addition decision trees add floating pop-ups with information about a node as the user hovers over a node. This is helpful when navigating large trees (because the nodes can get quite small).
Decision Simulator is an add-on to Blaze Advisor that is available with the 6.7 release. Integrated with both the IDE and with the web-based rule maintenance application is designed to allow the comparison of expected and historical data, results analysis and comparison of different approaches.
Like most things in Blaze Advisor, Decision Simulator is designed to be set up and configured by a technical user and then used by both technical and business users. The simulations are configured by identifying the projects that will be simulated (which can be in multiple repositories), the data they will use to run simulations and additional calculations that should be run on the output of the decision services being simulated. This last feature allows the business impact of decisions to be calculated and then included in the analysis. These formulas can be template-driven, allowing a business user to flex the exact calculation while running simulations.
Simulations can be run locally or, more usually, on a server and all runs are persisted for future and ongoing analysis. A range of out of the box reports are provided and can be configured. For instance, case detail reports can be configured for specific transactions, overall or by-segment statistics can be generated.
Once hooked up to rules repositories and data sources, business users can run and re-run simulations on all or part of the data and view a wide range of reports. They can also move results data into Excel for further analysis. Although there is no automated tie to information being logged from the rules engine the combination of the data loader for the Decision Simulator and the logging framework would allow this kind of direct feedback loop to be established.
While the overall environment is not as self-contained and pre-configured as some, the capabilities built into the underlying framework seem robust and extensive and I look forward to seeing what user achieve with it.
Blaze Advisor 6.7 is now the default version on their download site.
