I've been thinking lately about my ECM product and how to benefit from other technologies in Oracle's middleware, specifically BI.
BI has always been there for structured data, that resides in databases and structured systems, helping organizations to visualize what is happening in their environments, and suggesting how best to deal with that. Mostly, BI was targeting enterprise applications such as ERP and CRM, but in the real case, a huge amount of data is still unmanaged, which resides in unstructured data, such as emails, documents, sheets... etc.
those sources where very difficult to include in BI calculations, but since the introduction of ECM systems, the unstructured contents became available in a structured or semi-structured way, where metadata that describes business contents are stored in a database, as well as the full text capabilities available in most of current ECM systems. This created what we call a Content Enabled Vertical Applications (CEVAs).
Another important aspect here is the transactions targeting those business documents, which includes trends of users, types of requested documents, access methods, frequency of access, and other data that are collected and saved within the ECM system, those data are very important to build a complete vision of the organization's environment. Those transactions are directly related to the business and the decisions to be taken.
Let’s take a step back and look at the bigger picture in CEVAs environments, enterprise applications stores direct business records, and ECM systems stores content files related to those records, along with the descriptive metadata, fulltext, and the transactions related to the contents.
To analyze this heterogeneous data, a BI Engine is the choice to provide a comprehensive picture of the business and an effective BI driven decisions.
Oracle Universal Content Management
Oracle UCM can be easily used for this scenario, where all the metadata, and the full text is directly accessible through the database, and the content tracker component does a great job in collecting all transaction related data from the webserver filters and logs, the content server filters and logs and other sources, apply some cleansing on the data, and provide them in special tables to be accessed by Content Tracker Reports, and other reporting tools, and obviously, BI systems.
To make things more interesting, Oracle UCM has a component called Content Categorizer, that is a rule based auto categorization engine, which can fill metadata fields based on some text based rules, for example, assigning a content type according to the repeated occurrence of some keywords, or accessing the metadata in the actual file’s headers.
Those data can greatly help in the classifications of those contents, and give much more meaningful data to the BI engine.
Example of sales analysis
Let’s take an example of a CRM system that is used by a software company, this CRM stores information about customers, and the ECM system is used to store and manage the whitepapers about the sold products, and is also used to build the website that the public, partners, and customers are accessing.
When a whitepaper is viewed, the ECM system stores all related transactions, where the frequency of access to some whitepapers, or type of products can indicate the market trend, and the data in CRM can indicate the type of customers, or verticals interested in this whitepaper.
Enhancing the understanding of the market can help in determining the marketing approach, or even the development direction.
Example of Dynamic ADs..
A great example was discussed in the article about real time decisions, the example assumes that a user googled some search term, and one of the results landed on your WCM based site, where the pages are dynamically built, the system examines the search term used, consults its BI engine, and discovers the topics that are most likely to attract this visitor, and accordingly suggests an ad banner topic to be added, this happens very quickly while the page is being built, and the user sees the page with some AD that attracts him to click on!
The WCM system tracks if the user did click on the AD and notifies the BI system of that, or if the user didn't click on the AD which shows non-interest of the suggested AD. In both cases the feedback goes to the BI system to better tune the selection of the ADs.
In the end
If you are interested in this topic, I highly recommend that you visit the article here, it has much more information, and much more details.