As anyone doing business in Europe probably knows, the European Parliament adopted the General Data Protection Regulation (GDPR) just over a year ago in the Spring of 2016. The GDPR requires any company doing business in Europe to comply with strict new rules around protecting customer data. This has already introduced cause for concern among corporate security teams, as the GDPR takes a broad view of what constitutes personally identifiable information (PII). Companies will essentially need to provide the same level of protection for things like an individual’s IP address or cookie data as they do for name, address, and Social Security number. (more…)
As the data-driven age gives way to an information-driven economy where context is critical to surfacing useful insights from data, taking in relevance feedback from users, especially expert users, will play a major role in driving the benefits. This article explains the concept of a relevance feedback model and why you should care.
What is a Relevance Feedback Model?
Assume you ask a person or a system to provide you with information on a certain topic. There may be many facets to this topic, and you may get information from a whole range of different aspects. If you are working with that person or that system on a permanent basis, you may want to tell them that only certain aspects and hence certain kinds of information are relevant to you – in the hope of getting only the more relevant answers from them the next time you ask. You give the person or system “relevance feedback”.
Now let’s concentrate on a system, a cognitive information retrieval system or a “system of insight”. In that context, a relevance feedback model (RFM) is the capability of the system to take your relevance feedback and “internalize” it in order to tune the results of your future queries to what is most relevant.
The system performs and automates this task by adjusting weights attributed to certain terms and their equivalents (i.e. terms with the same or similar meanings) within the data it processes.
Imagine you asked “what do we have on MRO”, and you got information back on maintenance, repair and operations, but you told the system that you are only interested in anything pertaining to “Mars Reconnaissance Orbiter”. The next time you ask, you will get information only pertaining to the latter and possibly on related topics like Mars landing craft, automated robots for planetary exploration, etc.
For one person and one query, that seem rather simple. But now imagine, that you have tens of thousands of colleagues and thousands of topics to cover. That is when the RFM benefits from machine learning algorithms, not only to detect the preferences of each person but also of groups of people with similar interests, similarities in documents, etc. to spread the user relevance feedback to other documents, queries and people on an ongoing basis in an automated way.
Why use a Relevance Feedback Model?
A key benefit of a relevance feedback model is to enable users, in particular expert users, to affect relevance appropriate to their environment without the IT department having to implement rules for relevance according to specific user groups. It allows administrators to decide by configuration which specific users within the organization will contribute as well as the exact factor of relevancy improvement.
The relevance feedback model can also go a long way towards improving the human-machine interaction. As the relevance of certain content increases significantly due to relevance feedback, the user experience starts to feel much more “conversational” – i.e offering one to three suggestions as “answers” to a query – than a traditional search interface offering a list of documents in response to a query.
The RFM provides a way to discover from everyone’s experience the information that best answers the question. Take the real-world case of a customer service representative (CSR) seeking an answer to a customer’s product question using the product name or code. In this case, the CSR will obtain a diverse set of documents including parts catalogs, how-to information, product specifications, packaging information, marketing material, etc. All of this information is relevant but only some of it may help the CSR answer the customer’s question.
Thanks to the RFM, the CSR would immediately see information she has already viewed when she searched similar things in the past because the RFM takes into account the user’s “click actions” and applies a tiny relevance boost accordingly. Perhaps even more powerfully, the RFM will also modify the order of the results by observing (over time) what information other CSRs spend time to discover, even when they dive deeply into the results list for relevant information. Organizations striving to take full advantage of the RFM will configure it so that the experts’ interactions with the system provide bigger boosts for important content and even ban inaccurate information from appearing in results lists.
As you can see from the example above, the RFM provides a collaborative way to modify search result order. It is neither a tagging nor a classification approach, both of which can be done at indexing time (extracting metadata from source, entity extraction with Natural Language Processing) or afterwards (classification through ML algorithm like clustering, similarity computation, and so forth). The RFM arguably represents a smarter approach by directly incorporating human decisions when presenting information that will best address a user’s query.
As information-driven organizations strive for ever higher degrees of accuracy for end users seeking knowledge, the ability to leverage relevance feedback from users, especially expert users, automatically at scale becomes increasingly mission-critical for optimal business performance.
Forces of global competition, narrow margins, higher product development costs, and tenuous exclusivity holds drive organizations to push innovation, seek cost cutting strategies, and go-to-market as quickly as possible. Demands change frequently while regulatory and compliance standards become even more stringent. Organizations must keep up, and the pressure on research and development (R&D) never stops. R&D is the critical driver within the organization, whether within a large aircraft manufacturer or a leading automobile company looking to develop cutting edge products and services or a pharmaceutical company accelerating time-to-market for new drugs or a CPG company reinventing waning products. R&D thrives on information: customer information, expert information, product information, scientific information, market information, and competitive information.
To be at the forefront of innovation, R&D departments need complete visibility into both new and historical information across the entire enterprise as well as access to research from external public and premium information services. This is no simple task in today’s world where we are inundated with data — more data, more opportunities and more challenges. As a result, many companies depend on Cognitive Search and Analytics (CS&A) solutions to harness insightful, high-quality information and fuel innovation within their product and solution portfolios.
THE PRESSURE ON R&D
As organizations strive to create value, enhance customer experiences, and differentiate themselves from their competition, they have placed demands on their R&D departments to:
- Accelerate delivery of innovative products to market
- Optimize and manage available resources and knowledge while leveraging intellectual property
- Devise methods to reduce product development costs and eliminate re-work
- Improve product compliance both internally and externally and deliver safer, compliant products faster
- Understand consumer and market demands and improve responsiveness
- Resolve product issues quickly and efficiently to gain and keep customer trust
To meet these demands, R&D depends on complex scientific and engineering content that contains implicit conceptual relationships that can and should be semantically linked to simplify access to the knowledge embedded in that content.
HOW COGNITIVE SEARCH AND ANALYTICS HELPS
Cognitive Search and Analytics solutions amplify the expertise of R&D departments by surfacing insights from data across the enterprise, irrespective of location and format. From a single, secure access point, these solutions enable R&D professionals to unlock relevant and timely product research that helps make informed decisions. In addition, these capabilities are not limited to internal information; users can quickly access information from external Web sites and other applications, deriving relevant information and seamlessly integrating with internal enterprise information.
Cognitive Search and Analytics solutions enable enterprises to maximize the value of their intellectual property. Powerful search relevance and navigation capabilities enable researchers to find valuable pieces of past research and even parallel work going on without each group knowing about the other — eliminating duplicate work, reducing time spent in trials and shortening development cycles. These solutions allow employees to tag, bookmark and comment on documents, enabling collaboration and making teams more innovative, efficient and productive. Surfacing this existing knowledge enables workers to leverage the past work of distant or former researchers to benefit future research. Dynamically delivering relevant information, surfacing knowledge and enabling collaboration can decrease R&D costs significantly. Because R&D departments need to comply with a myriad of complex regulations, they need to be aware of relevant regulations without having to sift through the myriads themselves. This visibility enables R&D to stay abreast of regulatory mandates and efficiently manage compliance. Organizations can also leverage these solutions to send alerts to employees when there are new policy and compliance changes so that relevant R&D stakeholders are immediately notified.
Managing and maintaining product specifications is a critical function within R&D. Cognitive Search and Analytics solutions can access virtually any data source and expose changes when information is deleted or becomes outdated. These solutions can alert workers when any new information is created that impacts their specific process in the development cycle. These solutions also track and respect the access permissions accorded by each target application; only those with the correct privileges can access restricted information. Cognitive Search & Analytics solutions give researchers clear insight into product requirements and enable them to collaboratively develop safer, higher quality products that meet regulatory requirements.
RAPID RETRIEVAL OF RELEVANT INFORMATION MAKES THE DIFFERENCE
Extracting relevant information from vast and complex data volumes is a challenge that requires a sophisticated and scalable solution. The Sinequa Cognitive Search and Analytics platform handles all structured and unstructured data sources and uses Natural Language Processing (NLP), statistical analysis and Machine Learning (ML) to create an enriched “Logical Data Warehouse” (LDW). You can think of it as a repository of information about data and about relationships between data, people, concepts, etc. This LDW is optimized for performance in delivering rapid responses to users’ information needs. Users can ask questions in their native language or ask that relevant information be “pushed” to them in a timely fashion when it emerges. More than 150 connectors ready for use “out of the box” make the process of connecting multiple data sources fast and seamless. Company and industry-specific dictionaries and ontologies can be easily integrated, putting domain-specific knowledge “under the hood” of the Sinequa platform, making it an intelligent partner for anyone in search of relevant information.
The advanced semantic capabilities within Sinequa’s platform provide strong relevance in 21 different languages to assist organizations with even the most geographically and linguistically diverse workforce.
REAL-WORLD EXAMPLE: AMPLIFYING BIOPHARMA EXPERTISE
Consider one of Sinequa’s biopharma customers, a research-intensive organization dealing with a vast number of highly technical documents, produced both in-house and externally. The information in these documents varies according to the field of its origin – e.g. medical, pharmaceutical, biological, chemical, biochemical, genetic, etc. – and may deal with diseases, genes, drugs/active agents, and mechanisms of action. A lot of the information is textual, but there is also structured information, like molecular structures, formulae, curves, diagrams, etc. The volume of this information is on the order of magnitude of about 500 million documents and billions of database records.
Now consider the more than 10,000 R&D experts within the organization trying to leverage this information daily. They need to be able to ask topical questions, find relevant people and documents, and explore the vast information landscape to discover knowledge. The Sinequa platform supports this by plowing through the hundreds of millions of documents and equally large amounts of structured data, analyzing the data, analyzing the natural language user queries, and classifying results by category in real time. With the data tamed and enriched, it is presented to the user via a simple, intuitive interface with faceted navigation aids that allow the user to filter results further based on structural attributes that are either explicit or were intelligently derived by the system. The interfaces, also referred to as search-based applications (SBAs) are configured to expose functionality that is very specific to an R&D expert, aligning the solution with the goals of the user.
The Sinequa solution has proven to be very valuable to the customer in question, putting both internal and external research–related information that scientists need for research, development, and decision making into a single virtual repository with advanced navigation and retrieval capabilities. It has also proved to be very beneficial to teams of research and development contributors by allowing experts around the world to collaborate more easily through a single research application. Features such as navigation by topic across multiple repositories, de-duplication of similar documents, and improved research capabilities have all made knowledge workers more efficient and innovative.
Sinequa’s Cognitive Search & Analytics platform leverages relevant customer and market information to give R&D organizations insight and the ability to react quickly to demands. Teams utilize this platform to collaborate and share information. Sinequa effectively eliminates data silos and delivers relevant information from data to users in their business context, such that they can make better decisions, drive innovation, reduce risk, and be more efficient, which in turn enables forward-thinking R&D departments that thrive on continuous product improvements and introductions to amplify the collective expertise of the organization.
Mergers and acquisitions provide one avenue for organizations to grow via synergistic gains, strategic positioning and diversification. Even with an abundance of M&A activity, mergers tend to fail at the business process and information integration levels. The success of a merger can be greatly enhanced when business processes are integrated and information is seamlessly unified by gathering it from both organizations, analyzing it, establishing clusters of semantically similar information, and finding common patterns. Cognitive search & analytics platforms provide the necessary capabilities to accomplish all of this, thereby helping facilitate merger and acquisition initiatives and significantly increasing the odds of success.
ANATOMY OF A SUCCESSFUL MERGER
Let’s envision the details of how this impacts the relevant stakeholders. At the outset, a cognitive search & analytics platform provides the organization with unified access to information from both organizations and beyond. Users can leverage out of the box machine learning algorithms to explore and navigate this information. For example, the Clustering algorithm groups documents into topically-related clusters by analyzing the content and metadata. This is very useful for topical navigation and helps stakeholders identify similar documents based on named entities within the content. Automated classification is another useful technique for unifying information and improving navigability. In certain circumstances such as when classification rules do not exist but a properly classified sample set of content does, a Classification by Example algorithm can automatically create a model from the sample set, which can subsequently be applied across the combined set of content from both organizations to further enhance findability for stakeholders.
Machine learning algorithms can also help match experts with other experts as well as relevant documents across the consolidating organizations. This is done dynamically by analyzing what people write and collaborate around to compute user profiles, which are subsequently analyzed to compute “peer groups” that connect stakeholders with similar interests and expertise across the consolidating enterprise. With these peer groups established, experts can be more effectively presented with relevant content using a collaborative filtering technique that compares preferred content across the peer group and surfaces valuable content to members of the peer group who have not previously been exposed to it. As you can see, a cognitive search & analytics solution facilitates smart information sharing across the consolidating enterprise. Usually a lack of sophisticated security controls impedes greater openness between consolidating entities. A search-based application, however, respects existing security profiles—making it easier to merge infrastructures securely.
A cognitive search & analytics solution also helps to identify areas of risk and to solve outstanding issues before financial consequences occur. For example, risks could include content containing Personally Identifiable Information (PII) or content with no security associated. This is done by employing text-mining agents (TMAs), which provide out-of-the-box rules-based capabilities to extract elements from unstructured text. TMAs can be configured to incorporate terms and phrases specific to any part of the business. A cognitive search & analytics solution enables a quick, seamless and successful consolidation of organizations. Typically, in a large enterprise this is done as a series of search-based applications (SBAs) that each pull from a Logical Data Warehouse (LDW), which is essentially central cache of unified information. In the next sections, we will look at specific areas of the business that typically benefit the most from this approach.
SALES AND MARKETING
Once consolidation is underway, the organization must move quickly to combine sales and marketing activities, sales methodologies, pipelines and channels to drive revenue in the field and promote up-selling and cross-selling into new and existing market segments. The organization wants to minimize any potential lapse in the sales cycle for the newly merged company.
A cognitive search & analytics solution immediately equips sales teams with a single global access point to relevant, real-time and insightful information on products and customers—sales and customer notes, sales processes, product information and sales training are all immediately accessible. As previously mentioned, this is typically done using a dedicated search-based application (SBA).
An SBA for Sales and Marketing could provide unified access to the separate CRM systems and allow for the addition of shared content. As a result, no sales note, customer quirk or prospect is lost during consolidation. An SBA also offers the ability to push alerts and notifications out to users. As sales representatives learn about new products,
The SBA provides unified access to new marketing materials, sales process documentation, research documents and news articles to assist in training and ensures that they are effectively representing the newly expanded product line. Often when a merger occurs, customers know more about the products from the other company than the sales reps. With a cognitive search & analytics solution in place, an empowered and unified sales team can competently sell the acquired products and services.
FINANCE, ACCOUNTING, AND HUMAN RESOURCES
Finance, accounting and human resources are other key departments that need to be unified. A dedicated SBA provides a complete and consolidated access to information from the disparate ERP systems.
A cognitive search & analytics solution provides administrators and content curators with the visibility across the enterprise necessary to manage all documentation from both parties related to the merger effectively and securely.
Multiple IT systems and finance systems always increase the complexity of a merger. Organizations acquiring a large IT infrastructure need to identify the systems acquired and the value of the data in these systems.
A cognitive search & analytics solution enables the required visibility and helps to expose any redundant or unused systems that might be eliminated. For example, a cognitive search & analytics solution can be used to monitor and analyze usage of underlying repositories and applications to show what sources are being used less frequently. Some organizations have even been able to standardize and eliminate the need for multiple search applications. In other cases, a cognitive search & analytics solution works as a stop gap providing access to legacy systems that should be migrated, thereby reducing the cost of forcing an immediate migration.
Organizations can leverage Sinequa’s scalable platform as the cognitive search & analytics solution to connect information across the consolidating enterprise by leveraging previously mentioned machine learning algorithms like Clustering, Classification by Example, etc. as well as more traditional rules-based enrichment techniques. This helps the organization deliver unified access to stakeholders, ensure accuracy, reduce risk and gain insight into the complex systems across all affected departments. Enterprise Application Integration is usually a multi-year project. While cognitive search & analytics solutions do not replace such an integration, they provide comprehensive visibility in a very short time.
RESEARCH AND DEVELOPMENT
The cost of R&D increases when multiple teams unknowingly work on solutions to the same problems or fail to recognize and utilize the work done in past research. Cognitive search & analytics solutions positively impact R&D by accurately combining research data from the consolidating organizations, giving users real-time access and reducing the duplication of content, efforts, and sometimes entire projects. They do this by connecting experts working in the same subject area.
Sometimes a key driver in a merger or acquisition is gaining access to intellectual property, which often includes the expertise of the other organization’s knowledge workers. Sinequa’s Find the Expert capability gives employees from each organization the ability to discover the most knowledgeable people on a variety of topics, to view their profiles and to find associated information. This accelerates R&D discovery by enabling users to navigate information in clusters—leveraging the machine learning algorithm—and by content refinements.
These tools enable users to find past research and hidden relationships – including relationships with external experts with whom both companies have collaborated in the past – that would have otherwise been missed, thereby increasing speed-to-market.
The organization is also able to gain greater market share by leveraging and optimizing the information acquired instead of simply discarding projects in progress. One of the key drivers in a merger is the ability to retain and share as much knowledge as possible. Sinequa’s collaborative features spawn greater innovation in the newly expanded R&D department in the form of capabilities such as tagging, bookmarking and automatic feedback loops. For example, a scientist can comment on an old publication and explain how it relates to new research, which subsequently increases go-to-market speed by enabling broader collaboration.
HOW IT WORKS
Sinequa has developed an innovative and simple-to-use Cognitive Search and Analytics platform that offers Unified Information Access (i.e. information from any source, in any format, whether structured or unstructured, internal or external, through a single platform and user interface) to respond to even the most difficult information access challenges of large companies and organizations. The solution is composed of three main components:
- A powerful (natural language) analytics platform that gives structure to the most unstructured content. The analysis and indexing of data solve typical enterprise challenges involving multiple sources, unstructured and structured content, multiple languages, high data volume and high update frequencies.
- Simple and intuitive user interface that brings simplicity to complex search and analytics. Sinequa’s out-of-the-box user interface leverages the familiarity of common search tools enhanced with faceted search tools (i.e. filters) and analytics, giving the user an immediate picture of the information available from all enterprise sources. The UI is easily customizable to reflect specific appearance and navigation goals as well as corporate standards/branding, and is extensible, allowing third-party or custom UIs to be easily integrated via a provided search API (Application Programming Interface).
- A powerful, open and scalable technical architecture (GRID). Sinequa’s highly-scalable architecture was designed from the ground up to support multiple search needs starting on a single server cluster, providing a cost-effective solution that allows companies to respond to multiple business needs now and in the future with minimal hardware investment (whether the application is deployed on premise or in a private cloud). The architecture is distributed and modular to support virtually any data source addition and growth, often without any additional infrastructure investment.
Leveraging these combined components from within the Sinequa platform has proven to accelerate the integration of business processes and to amplify the expertise of the whole by seamlessly and securely unifying disparate information from each organization involved in a merger or acquisition.
A cognitive search & analytics platform facilitates information transparency and communication across the entire enterprise, minimizing disruptions while integrating teams and departments during mergers and acquisitions. Applying this technology as a solution effectively operationalizes the information access, speed and control needed to ensure long-term success in the newly formed organization.