Psycho-cognitive search- A steroid for KM?

France continues to incubate search, content processing and analytics innovations. Exalead, founded by a member of the Alta Vista search team, challenged Google with 64-bit architecture in the early 2000s. Other notable French companies offering a range of capable information retrieval solutions include Antidot and Sinequa.

In my experience, systems developed and nurtured in France have three distinctive characteristics. First, the companies implement specific philosophical ideas in their systems. Sinequa, for instance, takes an integrative approach to information access. The original Exalead embraced the mathematical foundation of France's elite universities.

Second, the solutions handle French language content better than search systems crafted to address findability problems in English or Chinese. The French language  poses formidable grammatical challenges and fast-changing expressive idioms.

Third, French systems, like many facets of the country's culture, possess a je ne sais pas. I'm not sure what makes the French approach refreshing from an engineering point of view. I do know that I am delighted when I discover a new system with an interesting name and intriguing approach to findability.

On a recent trip to Europe, a colleague mentioned Oorace, a system developed by Search'XPR, headquartered in Lyon, France. Founded in 2011, Search'XPR has an impressive pedigree. Co-founder and CEO Jean-Luc Marini is a well-known professor in a country that puts significant emphasis on intellectual achievement. The other co-founder is Olivier Fignon, who has both academic and business experience. The two teamed to commercialize psycho-cognitive methods for information access.

Subconscious influence

Oorace is not a ready-to-run search system like Google's Search Appliance or the enterprise version of LucidWorks. Oorace is an application programming interface (API) for information access. The results generated are a semantic map that echoes Claude Vogel's Semio relationship map and Inxight Software's hyperbolic map. Relationship and semantic maps have become increasingly important to those who want to get a visual picture of how search results fit together.

The company describes Oorace as a "revolutionary API that greatly enhances traditional applications for data search and recommendation. The foundation of Oorace is a proprietary algorithm that explodes the results of traditional keyword and semantic tools. It incorporates ‘psycho-cognitive' principles that utilize the way people subconsciously acquire, process and store information."

The firm's website states: "The psycho-cognitive principle focuses on the influence of the subconscious on the data search. This principle is based on analysis of user behavior strategies and capitalizes on past search experiences and not on the meaning of the words or semantics. It constructs and provides search results that appear to be unexpected, but are actually linked to a person's subconscious. This outcome is often described as "serendipity."

Serendipity

Intrigued, I contacted Search'XPR, and spoke with Jean-Luc Marini about the company's approach, particularly the psycho-cognitive angle. Search' XPR's method refers to "an unexpected recommendation—a search for content, a person or an organization within a digital memory." According to Marini, "This recommendation, with no obvious connection with the initial search, is touching on one of the user's topics of interest about which he had no conscious awareness." The company uses the word "serendipity" to describe that suggested relationship presented to the person running a query.

Marini said, "The implementation of this principle opens a way for a new generation of applications embedding the recommendation technology developed by Search'XPR: the Serendipity Apps. The technology is disruptive, since it provides additional functionality to existing recommendation technologies."

As an application programming interface, Search'XPR is unlike a one-size-fits-all search system such as HP Autonomy or Microsoft SharePoint Search. Via the API, a developer can tap into the functionality of Oorace. Marini explained, "Oorace APIs are, for example, used to integrate the serendipity feature into TV listings and VOD (video on demand) search and recommendation systems connected to the Internet. The feature is gaining leverage from the user's behavior and his subconscious topics of interest not only linked to his TV listings and VOD searches but also linked to his searches and navigation on other online services offered by the operator of the Internet set top box.

"Because of this, the operator is able to offer the customer, during his navigation, some totally unexpected listings or VOD compared to its search topics, but totally linked to his subconscious topics of interest. In the future, the associated ad agency will also be able to use this feature to offer in the same way unexpected and well-targeted ads to the subscriber."

Managing different data

The system uses a representational state transfer (REST) protocol in the cloud. The Search'XPR psycho-cognitive functions and other features are available to the licensee. The operation of existing enterprise software is not affected by the Oorace functionality.

In many organizations, content is scattered across different servers. Some content may reside on the public Web in the form of pages on an organization's website. Other information may be available from third-party providers. Search'XPR's system processes information available to the licensee.

Marini said, "From our perspective, because we are becoming an addition to the current data search systems, the data storage problem does not speak the same way. Search'XPR's technology requires managing different types of data. Data we are managing within our recommendation technology are not only structured data but also graphs for which we manage evolution over time. Therefore, our current storage architecture relies not only on a distributed non-relational data management system with a structured storage for large tables, but also on an object-oriented database using graph theory adapted for the use of graph type data. In addition, Search'XPR is currently considering organizing a new model of dynamic data and graphs storage in a distributed environment."