Technology for Everyone


Discovix has developed a content discovery and text analytics technology that combines sophisticated mathematical models and machine learning techniques to identify important information within massive data sets.  Although Discovix’s technology is highly advanced, it is a simple and intuitive system to use.  Discovix provides an advanced Curiosity Engine, which guides users to relevant content and makes topic-based recommendations once they train the system through supervised learning techniques.


Context, Expectations, and Knowledge

The DISCOVIX Curiosity Engine is built to leverage user expertise. Human intuition is more advanced than a computer, and software should reflect that, not mask it. In addition to subject matter expertise, human users are endowed with the power of situational awareness and instinct.  DISCOVIX acknowledges that anomalies are the most important types of information in data sets. Anomalies are violations of expectations. Expectations can only exist in the presence of context, and context evolves with knowledge. For example, when analyzing a demographic data set of population height, a man who is 6”1 may seem exceptionally tall. However, if it is revealed that the population is NBA players, the context and expectations change. A player who is 6”1 is no longer exceptional.


DISCOVIX proactively seeks relevant information for the user from the dataset. Using unsupervised learning techniques, DISCOVIX presents the user with an initial queue of sample documents. Through vectorization and clustering, the DISCOVIX system is able to quickly evaluate the corpus of documents, identify topics and correlations, and present them to the user. Once the user is presented with data, he or she can label the data, thus beginning the supervised learning process and imparting their expertise to the machine. As the machine learns, it is able to mimic the user’s decision-making process. Using the initial queue of labeled documents, DISCOVIX can discern what information is and is not important with the intuition of a human and at the speed of a machine. Ultimately, the computer will learn enough from the user that the user can stop training the system, and wait for the machine to present relevant data. Imagine the potential of your mind at computer speed.


What is A Curiosity Engine?

Anyone who’s ever conducted online research is familiar with the problem of excessive information. On- line users often encounter insurmountable amounts of data during research. This overabundance of data is overwhelming and impedes the discovery process. Isolating the most important pieces of information is a tedious task made inefficient by the current keyword search paradigm.  Keyword search does not encompass the complexity of research and discovery. The curiosity engine is the evolution of search. Rather than fetching dictated pieces of information, like keyword search, a curiosity engine proactively seeks topical information.


The difference between the keyword search paradigm and a curiosity engine is that keyword search is mere content retrieval, while a curiosity engine enables in-depth discovery of BOTH content and context. In essence, a conventional search engine acts as a servant, bringing the user only what is specifically requested. This oversimplifies the research process. Researchers explore complex topics, attempting to isolate a thesis or solution that is not easily identifiable or defendable. !e ‘answer’ does not lie in a single piece of data it is found in a story composed of pieces of a data set. A curiosity engine acknowledges the complexity of the ‘answer’.  Acting as a butler, not a servant, the engine anticipates the needs of the user before the user does. The Curiosity Engine preemptively locates germane information and presents the user with a queue of data, ranked in order of their relevance to the topic. As the research process progresses, the results becomes more precise.


In order to anticipate the needs of the user and evolve with the discovery process, a curiosity engine must receive feedback from the user. The curiosity engine simply asks that the user input his/her opinion in the form of ‘yes’ or ‘no’ for each document read. This process enables the system to learn from the intelligent user, so it can rapidly vet data and only present the most important information. Using mathematical reasoning and an innovative set of machine learning algorithms, the patented DISCOVIX Curiosity Engine inherits user knowledge. With a small amount of user input, DISCOVIX applies human intuition at computer speed.


When the research problem is more complex than looking up directions or fact checking, a curiosity engine is the only tool that should be used. Companies with large, multifaceted problems cannot expect keyword search to do anything but slow down the research process and overwhelm the user with the burden of search. A curiosity engine allows for modifications and adapts with the user as the question and its resolution become clearer. The DISCOVIX Curiosity Engine is the most efficient tool in terms of locating, analyzing, and organizing information in massive, structured and unstructured data sets.