Towards a business-aware framework for the development of smart process applications

点击翻译按钮获取中文摘要

You walk into your favorite grocery store, and you are greeted by name on a major display as you go through the door. You walk by the meat section, and a text message (or head-up display) brings to your attention a special on lamb chops because your significant other is known to relish those. You put a meat tray in your shopping cart, and you get two thumbs up on Syrah wine. Cheese with that? You pick crackers, and get a warning about sodium content, given a medical history of high blood pressure. While getting toothpaste, you get notified of a special on size 4 baby diapers, considering that you have been buying size 3 diapers for over 6 months.". The scenario is an example of a Customer Experience Management application in action that relies on connected objects, knowledge about the customer (facts and inferences), and an implicit understanding of the purchasing process  to trigger custom interactions with the consumer at the opportune time to achieve a desired business outcome. Consider now a patient monitoring application that relies on sensors attached to patients or their environment and that, depending on prior knowledge about the patients and their surroundings, 1) notifies, acts, or reacts, if need be and 2) records and "digests" the data to refine future responses. Or an insurance application that relies on sensors to capture a driver's driving events to 1) notify, act, or react, and 2) record and "digest" the data to refine future responses-and perhaps raise the premium! These are three examples of applications that 1) use ubiquitous computing for context-awareness, 2) rely on elaborate subject profiles, which include factual data and domain inferences, to compose a response, 3) have a learning component that adapts the application's response over time, and 4) that implement novel business processes to achieve a high-level business outcome, e.g. more sales, happier customers, less risk, etc. I will refer to these as smart process applications or SPAs. SPAs raises issues, from business models, to application engineering, to data privacy and security, to legal and ethical issues. I focus on application engineering. Whereas traditional business applications implement well-understood processes, such as supply chain or finance, SPAs implement dynamic novel processes that often instrument or accompany human processes that are outside the scope of the system. The way industry approaches this is "let us put sensors everywhere, collect the data and see what we can do with it". However, there is no systematic way of specifying requirements for SPAs. I argue that we must first codify the processes that we wish to instrument before we can specify what the SPA will do. Thus, I propose to develop a requirements methodology for SPAs that relies on an explicit codification of the underlying processes. Second, I propose to leverage the codification of such processes and similarities between SPAs to develop the domain functionality through frameworks.