How to Build a Business Rules Engine - Advanced

[Rx_]

Does anyone have recommendations for developing a Business Rule Engine?

In essence, it looks at various records and provides a seeming intelligence guide to the user.

The license for a .NET single workstation on one product was $35,000 USD. Some are even more. Biz Talk's rule tool - it can work for accounting and inventory. Not that great for more complex decision making.

One database manages Legal, Regulatory, and Policy for permitting. The Regulatory covers Federal, State, Local and many other such as Native Lands, Bureau of Land Management, Parks... It gets really complicated.

The idea is to build a group of functions that play together.
For a record or change in a record - it evaluates several dozen tables. Makes a judgement. Kind of like a tax form - when some conditions and values exist, you should create and file a different form. Or, you should file as single instead of joint. Or both.

The best I found for under $20,000 (so far).
Amazon has this book (I have yet to receive it - if you have, please comment):
How to Build a Business Rules Engine: Extending Application Functionality through Metadata Engineering
The reviews for concepts are good. The C++ Programmer complained that the example DB uses MSACCESS and the code is in VBA.
NOT a Recommendation:
Review this at: http://www.amazon.com/Build-Business-Rules-Engine-Functionality/dp/1558609180

 

[spikepl]

The license for a .NET single workstation on one product was $35,000 USD.

Have you thought about why the price is as it is?

I once worked on a huge project intended to provide a decision-support system for management, for handling operations, renewal and maintenance of a large complex mechanical infrastructure system. With regular recording of the the state of the mechanical system and the load on it, and with a crystal-ball function to be designed by us, the system was among other things meant to predict future maintenance/renewal needs. 2½ years and about £ 2.5 million later, this veritable miracle tool was scrapped. There were of course many reasons for that, but one very prominent one was the optimistic expectation that an extremely complex system could be modelled and thus reduced to bits and bytes, and that the system could be made clever enough to replace a lot of human judgement concerning logically very fuzzy issues.

In conclusion: Business Rules engines (and their application), AI, decision support systems and, in general, systems of that ilk are not simple things. Unless you work in some R&D organisation, then I'd recommend preparing and reviewing a business case. List the various options with expected benefits and expected effort (for development, customisation, testing, verification and implementation of an in-house, or externally-acquired system) prior to dabbling with stuff like this.

 

[Rx_]

My exact thoughts, with many other experiences that mirror yours.
You deserve a THANKS because both your experience and honest are welcome!

Years ago, I took the Biz Talk course. That worked great across the hallway where the decision tree process was rather elementary. Things like applying for your first or retrieving your birth certificate.
But, in the regulatory side, the government spent about $millions on a Sharepoint application with all the fancy rule development. That product was never used a single day. The code maintenance cost was way too high due to complexity.
I am sorry to tell taxpayers, that was just one of many experiences.

Meanwhile, somehow the Access programmers just kept turning out things that worked and never got credit.

UPDATE:
I bought the book on Kindle. After a bit, was able to download the bizrules.mdb from the associated web site. Have only 10 minutes to review the code. The code is professional. It is well documented. This code is not expected to be the "end-all". But, if the book, tables and code give me some good structure?
Will let you know soon.

 

[Pat Hartman]

I have also worked on mega-mainframe projects to do stuff like this. The projects take many man-years(If one woman can make a baby in nine months, then obviously nine women can make that same baby in one month. At least that's the way the pointy haired boss' think.), cost millions, and have a high rate of failure.

$20,000 per seat is probably a bargain in the long run especially given you get to start using the system now rather than 2015 (maybe).

 

[Lightwave]

Be cautious in your business activities because the world is full of trickery.

Are you buying an application or just the environment?

I wouldn't pay that for an environment as we already have.

MS Access
Filemaker Pro
SQL / VBA / C++ and every other programming languages and IDE out there.
I consider these all to be the equivalent of business rules engines.

Now if someone could teach me how I could create my own programming language now that would be an interesting concept. Look at the value of SQL it is worth far more than that kind of money.

For an application I guess it depends on the value of the process and information contained as much as it's intrinsic "complexity"

I could see a situation where I would be happy to pay a chief executive to hand deliver and cash a cheque if that cheque was of sufficient value.

 

[jdraw]

I have also worked on some mega government projects - some were successful, others, as Pat has noted, conceived by the pointy haired at the 30,000 ft level, and left to others to "fill in the details".

You might want to look at Business Process Management Software which may be more encompassing than business rules creation. I haven't used any products/packages, but I am aware some exist, and some may be head and shoulders above others.

Getting business rules defined for any project can be a difficult task. But getting rules that are stable for businesses that are complex in organization; varying management styles; subject to staff and management changes; changes in policy decisions and with multi-site presence makes any longevity of design fleeting at best. Just getting different groups to agree on the definition of Address (let alone representation within computer systems) can be a chore.

I would not be putting a lot of faith into a all encompassing business rule engine at this time. I sort of come from the do it in manageable-chunks group - that's not the 30,000 feet level and someone else fills in the detail.

 

[spikepl]

Inspired by jdraw, I'd suggest that you have a quick play with Bizagi process modeler - www.bizagi.com - easy to use (on the surface very much like a flow diagram) and yet can get pretty damn complex.

 

[Rx_]

Thank you everyone for your input. Amazing how much in common we all have on this subject. To be sure, there is not a straight-forward answer. And, I will never get $30K funding for a tool that might not fit my exact situation.

The following is just an opinion and status report:
http://www.bizrulesengine.com/index.cfm?fuseaction=publications&mode=books
This is about the book and the download site for the Access DB used as the example in the book. There are some free interesting articles here as well.

What lead me to spending a whopping $65.00? One person gave it a good review on the "how to", but complained that the Sample DB used in the book was written in MS Access. He expected code in C++.

The writer has credentials that looked good. Other books on more esoteric high-level design theory.

This book appeared to get a good review for the essence of developing your own Business Rules Engine.

Access Download so far looks well designed and commented, all using VBA code. It will take some time to fully evaluate it.

This book is not intended to solve the problem. It is more about having a DB to organize and gather the business rules and propose a methodology to apply them.
If it is a better template methodology for what I would end up doing anyway, it might be worthwhile. If not, I won't hesitate to let everyone know.

 

[Lightwave]

You might want to take a look at something like this

Open Source Diagramming software

This looks quite good as well

flyinglogic

It appears to allow some dynamic logic setting which is not something I've come across before.

 

[Rx_]

Book Review Lite:
Forward: ".. his build-it-yourself approach is not for everyone. For a more generalized rule environment, or for inference-oriented problems, a commercial rule engine is generally a better choice. But business rules are literally everywhere in operational business processes, and there are huge opportunities for rule-oriented development on a smaller scale.

A major reason for building a business rules engine is cost. Commercial business rules engines can cost hundreds of thousands of dollars to purchase or license. Even after that, they must be integrated into the enterprise's IT environment. In particular, they need to be configured to acquire the data they need to process. This can add significantly to the overall cost. They are usually not pure rules engines, but include components that are unique to the business area they are addressing, such as a module to build a Chart of Accounts in the case of accounting packages. Furthermore, even if integration is achieved at one point in time, there may be a cost to ensuring that it is maintained over time.
How Easy Is It to Build a Rules Engine?
To build a generic rules engine is extremely difficult, and even the commercial products available tend to address different needs. None of them is truly generic, and they all fill different market niches. To build a business rules engine that extends a particular application or addresses a set of specific business requirements should always be much easier. This book provides a methodology that can be followed for developing such a rules engine. Designers are encouraged to apply this methodology to their sets of requirements and then estimate the effort required in their particular situations.