Combining traditional enterprise data management best practices with emerging semantic approaches and technologies helps the organizations to become more agile and effective in a dynamic business environment.
Cindy Walker spoke at this year’s Enterprise Data World 2014 Conference about using the semantic approaches to augment the traditional enterprise data management practices.
She talked about the benefits of semantic approaches and the value of a combined strategy of integrating the traditional data management with semantic technologies.
In the presentation, she also discussed the phases of adoption of this strategy. Cindy also led the creation of a Center of Excellence (CoE) for data analytics.
InfoQ spoke with her about the data management best practices and the data analytics center of excellence initiative.
InfoQ: Can you give us an example of how semantics can help enhance business agility in organizations?
Cindy: Sure. One example is the inferencing capability of semantic technology. Inferencing can reveal context and meaning (e.g., relationships, categorizations) that enable the business to make better use of the data right away – to interpret, analyze, and act quickly - without having to wait on the data to be modeled and structured (as with relational technology) or application code to be written (as with XML) to reveal the meaning.
InfoQ: What are the emerging trends in the field of semantics?
Cindy: There are several trends taking place in the field of semantics. One trend is the industry-focused collaboration efforts to develop and link domain-specific standard ontologies, such as the Financial Industry Business Ontology (FIBO), to facilitate information sharing and regulatory reporting. Use of standard ontologies can help organizations connect disparate data sets and can enable semantic queries across the web and data sharing with internal and external stakeholders easily (without restructuring the data or developing point to point interfaces.) FIBO is being developed in phases by volunteer contributors in the financial services industry (with some financial regulator participation) under the authority of the Object Management Group. Their goal is to use the FIBO so that data can be shared easily between bank divisions, institutions and regulators—as a means to better understand and manage big-bank risk profiles.
Another trend I’m observing is a growing interest in exploring the use of semantics within individual organizations to facilitate easier, faster, and cheaper data collection and integration into enterprise data warehouses or into master data management systems. Operational ontologies and triple store data structures can help organizations link disparate data sets by using triple stores and graph data structures that can be more efficient than relational data structures to address certain business and technical challenges.
An additional trend is the potential use of semantic rules-engine-type technologies to monitor business transactions and enforce compliance with an established set of rules (such as mandatory laws and regulations.) These semantic rules engines can, for example, help exploit the power of the ontology to translate and ingest text-based legal statements, into active rules that are enforceable near real time and the triple store data structures to interoperate across disparate applications.
InfoQ: You also worked on establishing a Center of Excellence for data analytics. Who do you see should be part of the center of excellence team in Business and IT groups?
Cindy: I see Business Analysts and Subject Matter Experts as key members of a Data Analytics Center of Excellence team. From the IT group, application developers and DB Administrators are essential. These are the team members who guide the development of the SQL and NoSQL databases and write the ETL programs in the relational environment and in the Hadoop ecosystem. Data scientists, of course, or data science teams, are on the team to design, develop, and test the predictive models. And data visualization specialists provide the guidance to ensure that the visual presentations of analytical results are optimized to make it easy for the users to interpret and use. I also see ontologists and data architects as vital members of the Data Analytics Center of Excellence team to translate business needs into the conceptual designs and semantic layer are foundational to the development of our analytics solutions.
InfoQ: What is the ideal size of the team?
Cindy: In my experience I have observed that many factors influence the ideal size of the data analytics team. For example, scope of the problem and number and complexity of the predictive (or prescriptive) models required to address the business needs. Also, number and complexity of data sources to be analyzed impacts the team size. And, of course, the level of business importance in terms of time-sensitivity can impact the size of the team in some cases.
I have organized Salient’s DACoE team into three primary focus areas from which we can combine resources to work seamlessly together on high performing teams. These focus areas include Agile Business Intelligence, Big Data Management, and Predictive Analytics. Each focus area will have a dedicated Director and a team of individuals supporting the activities in that focus area. For example, the Big Data Management focus area team will have developers who work with NoSQL databases as well as tool specialists who work with Open Source tools, such as Hive and Mahout. The size of each focus area team is flexible to meet our clients’ needs.
InfoQ: What are the constraints or challenges in establishing a Data Analytics Center of Excellence?
Cindy: One challenge in establishing the Data Analytics Center of Excellence could be the limited availability of formally trained data scientists. I‘ve taken steps to overcome this challenge by reaching out to universities who are offering data scientist education programs and by establishing teams of people whose collective skillsets enable them to perform data science functions, such as predictive model development, hypothesis development, quantitative analysis, etc.
InfoQ: How do you measure the success of a Data Analytics Center of Excellence team?
Cindy: By measuring the business impact of the insights gained through the use of the analytics results. For example, ultimately our goal is to make the organization more agile – that is, to help them make better, faster decisions, and more importantly, to take timely, effective actions that make our clients stronger, more competitive, more profitable, or (in the case of public sector clients) more effective at performing the core mission. So, I’m leading my Data Analytics Center of Excellence team to develop explicit metrics that track both leading indicator metrics and lagging – or outcome – measures. We want to empirically demonstrate (with “before” and “after” performance snapshots) that the analytics we’re providing are making a positive impact on the business. For example, we measure the percentage of strategic goals with key performance indicators that are supported by analytics visualizations.
About the Interviewee
Cynthia (Cindy) Walker is the Director of Salient’s Data Analytics Center of Excellence. She leads the developing and implementing Salient’s Data Analytics strategy and expanding market distinctions and innovations of the DACoE to include predictive analytics, agile business intelligence, and big data solutions for Salient’s clients. Ms. Walker has over 25 years of information management and business intelligence expertise and has served as strategic advisor and architect guiding enterprise transformation, knowledge management, enterprise data management, and information sharing efforts across the Federal market.