Intelligent understanding of content use
Content creation and retrieval capabilities continue to improve, but for knowledge management, a gap clearly remains: How do we know what’s actually useful—and being used? That is critical in enterprise environments where “content” is often an array of files, documents, data and Web pages. Relevant insights are not measured in whole documents, but by what’s found inside, embedded in concepts and illustrations. While technology can semantically model concepts within content, and basic Web analytics can identify document “hits,” recognizing value comes from what people do with the information.
Questions at the heart of cognitive computing
How do we know what people find valuable in available content and how they apply it in their work? How can we use those insights to intelligently support other people? And how can we assess what effect relevant content has on organizational performance outcomes?
Those are questions at the heart of cognitive computing, within and beyond the context of knowledge management. Cognitive application design starts with understanding the value of specific types of content relative to a user’s decision context and information needs. It is based on work patterns and question-answering expectations—enabling the machine to act as guide and support rather than as task master. In order for cognitive computing to fulfill those expectations, we need to vastly improve the transparency and availability of critical information and its contexts. For that to happen, we must understand the roles and contributions of governance in our information practice.
Organizational governance requires aligning resources—in this case knowledge resources—with customer and internal expectations in order to deliver systemic value. Content governance requires understanding information needs and guiding content producers to create and sustain information value over time. Those areas of governance require clear insights into the use of information and its impact on the organization and customers/stakeholders.
Gathering information from content use
Our analytical tools need rich, real-world inputs. Leveraging a different approach to managing people’s interactions with search results and content provides that input. I described an annotation-based approach to managing search and content use in “Understanding Content Use in Enterprise Settings” (designforcontext.com/insights/understanding-content-use). The approach captures data on content selection at very granular levels (down to sentences or images, with each individual’s organizing schemes and context), and then facilitates the way people apply information in their work. Insights can be drawn from that approach and applied to ongoing organizational and content governance, with the support of cognitive computing.
Enabling systems to capture and analyze the use of concepts, with its surrounding context, helps us understand:
- how people conceptualize what they need, creating data about how searches lead to specific document and concept selections;
- the iterative evaluation processes people perform on resources over time, for example, over the life of a case or when preparing research reports;
- how people learn through their use of content over time;
- how resources and learned concepts are applied in work products, job outcomes or shared among colleagues; and
- the lifecycle of usefulness of specific content and concepts.
By capturing an array of detailed data points on content selection and use, then aggregating it over thousands of people and many thousands of searches, we can apply cognitive computing to learn the patterns behind concepts applied in specific organizational contexts. The patterns then provide feedback to improve relevance, recommendation, proactive support and decision modeling—a feedback loop that learns over time and informs governance practices. We understand how content turns into action.
How can we use insights for governance?
- identify key content areas and gaps in available information;
- spot unrecognized redundancies where concepts are “reinvented” in different parts of the organization;
- analyze trends in information needs to align with customer, market and societal expectations;
- prune or refine content to align with a new context; and
- identify good examples for authors—in context of what they are creating.
- identify relevant, yet missing, tagging for document metadata;
- support taxonomy maintenance when analysis indicates terminology “drift” or new emerging concepts;
- target when, and where, to adjust categorization schemes;
- feed real-world usage patterns into automated classifiers to intelligently map adjustments back into content assets; and
- elaborate organizational context models, allowing them to be applied to other uses.
- enhance work product quality assessment by understanding reference sources,
- plan targeted training and performance improvement programs,
- align available resources with your risk management practices, and
- foster collaborations for innovation.
Cognitive computing can play a key role in clarifying patterns that provide insights for governance. At the same time, cognitive applications are beneficiaries of sustainable content governance foundations. This all starts with capturing detailed data related to contexts of use. Taking a richer approach to data gathering allows you to make well-informed improvements over time. This, in turn, sustains a high quality of information available to your users and to your cognitive computing applications.