A decision engine is a type of web-based computer application that attempts to aid the user in arriving at a decision in one of several ways. A common case use is online shopping, where a customer inputs his priorities for a given product and the decision engine determines which particular makes and models most closely fits his preferences. Decision engines can also work by tracking a user's searches over time and using the collected data to make suggestions.
Decision engines should not be confused with search engines. Search engines are a centralized location from which to access different information. Decision engines, on the other hand, produce individualized search results based on a number of criteria.
Instead of being a base from where a user can run searches, a traditional decision engine model is intended to have its topics return as search results by other search engines. For example, a user may type a question into a search engine. One of the top results for this search would be a relevant topic on a decision engine.
Once on a decision engine, a user is presented with a series of questions, known as a decision tree, designed to eliminate choices on the path to finding the most ideal option. If a user searched for mobile phones, likely questions would relate to price, size, carrier, and desire for options such as speakerphone, web-capability and so on. Based on the answers to such questions, the highest ranked answer is ultimately presented with an accompanying explanation.
One of the main drawbacks of this decision engine model is that topics must be created before they can be used. Similar to a wiki approach, such decision engines require user participation and are dependent on community development to become more effective. Decision engines that rely on human input are also similarly subject to human subjectivity and opinion.
A common solution to bias in decision trees is to enable community voting. The best or least subjective entries rise to the top, while poorer entries are buried. The reliability of voting to quash poor entries also improves with more community involvement, making it even more critical to have a large and active user base.
More automated decision engine models are incorporated into popular search engines and work on the basis of using accrued search data to suggest results the user is likely to find useful. Instead of relying on human input, these recommendations are produced on the fly according to predetermined formulas. Users can improve results by telling the system whether or not they are helpful.