Archive | Metadata RSS feed for this section

I Never Metadata I Did Not Like

30 May

This presentation is from December 2019 – pre pandemic, and is the last live show I have presented in – I do look forward to getting back into that groove. The following is a write up done by Amber Dennis at DataVersity. Also posted on LLRX.com – here.

Managing Metadata: An Examination of Successful Approaches – DATAVERSITY

Managing Metadata: An Examination of Successful Approaches

By Amber Lee Dennis, 30 Nov 2020

“If Google can deliver results across the entire internet in seconds, why do I have so much trouble finding things in my organization?” asked Jonathan Adams, Research Director at Infogix, at the DATAVERSITY® DGVision Conference, December 2019. In a presentation titled, “I Never Metadata I did Not Like” Adams outlined successful approaches to understanding and managing metadata.

What is Metadata?

According to the DAMA International Data Management Body of Knowledge (DAMA-DMBoK2), the common definition for metadata, ‘data about data,’ is too simple. Similar to the concept of the card catalog in a library, metadata includes information about technical and business processes, data rules and constraints, and logical and physical data structures. It describes the data itself, the concepts the data represents, and the relationships between the data and concepts. To understand metadata’s purpose, imagine a large library, with hundreds of thousands of books and magazines, but no card catalog. Without the card catalog, finding a specific book in the library would be difficult, if not impossible. An organization without metadata is like a library without a card catalog.

“Obviously it’s data about data, in that sense. We all know that, but also, one person’s data is another person’s metadata. So it gets kind of confusing,” Adams said. Metadata has traditionally focused on technical metadata, which details the structure of data and where it resides, supports IT in managing data, and assists user communities in accessing and integrating data. Reference data, which provides known vocabulary and creates business and operational context along with semantic meaning, is also metadata. Adams said:

“Metadata is kind of everything. It’s how you visualize it, and it’s how you find it. It totally enables data, and in many respects, it’s going to be the bulk of the data you have.”

Types of Metadata

  • Descriptive metadata is metadata about the asset, including its title, creator, subject, source, keywords, etc.
  • Content classification metadata details the content and meaning of the data asset. This includes relationships, data models, entities, the business glossary, controlled vocabularies, taxonomies and ontologies.
  • Administrative metadata details how to access and use data assets and includes lineage, structure, audit and control, and preservation information.
  • Usage metadata indicates how data may be used and how it must be controlled, which includes users, rights, confidentiality and sensitivity.

 “And if that isn’t complicated enough,” he said, “those four types of metadata get applied slightly differently depending on where you are.”

Metadata for Operational Systems

Adams provided an illustration of an operational system using a pyramid, with reports on the top level, transactional data on the second level, then functional data on level three, master data on level four, and structural and reference data as the base of the pyramid. Types of data not included in this structure might be a data lake used by marketing, external data, financial information, or CRM data:

“This gets complicated, so we’re going to talk about simplifying it. My point here is that you should drive it from the user perspective, with that use case, view it within this context, and scope it appropriately.”

How Is Metadata Important?

Metadata answers critical questions about data:

  • Is the data discoverable?
  • Is it understandable?
  • Can it be accessed?
  • Is it usable?

Success in Metadata Management is shown by how well a team engages and aligns information to the business and operational context of the organization, Adams said. The DMBoK2 says that like other data, metadata requires management. As the capacity of organizations to collect and store increases, the role of metadata management grows in importance. To be data-drive, and organization must be metadata-driven.

Success with Metadata Management

To manage metadata, start with a framework that aligns data to business and operational contexts so that metadata can support Data Governance in the following areas:

  • Organizational Impact
  • Capabilities and Interfaces
  • Programs and Platforms
  • Repositories

Adams then further broke down how to address the governance of each of these four areas.

Organizational Impact

Metadata turns critical ‘data’ into critical ‘information.’ Critical information is data + metadata that feeds Key Performance Indicators (KPIs). He recommends asking: “What will change with a better understanding of
your data?” Getting people on board involves understanding how metadata can solve problems for end users while meeting company objectives. “We want to be in a position to say, ‘I do this and your life gets better.’” To have a greater impact, he said, avoid ‘data speak’ and engage with language that the business understands. For example, the business won’t ask for a ‘glossary.’ Instead they will ask for ‘a single view of the customer, integrated and aligned across business units.’ An added benefit of using accessible language is being perceived as helpful, rather than being seen as adding to the workload.

Capabilities and Interfaces

All users must be given the capability to discover information and apply it to challenges, to share critical information, and have access to automated process when available.

  • Discover and Understand: A catalog search portal allows users to discover what data is available, place that data in context, and understand who can access it, and how to do so.
  • Communicate and Share: Users need the ability to communicate what they’ve produced and make it available for broader consumption. Complete descriptions of data are necessary for compliance and consistency, but must be available in language geared toward the user. The term ‘ETL processing’ may be adequate for an IT user, but terminology such as ‘GDPR compliance’ should also be available so business users have access to the same information.
  • Acquire and Integrate: Acquisition and integration varies depending on the perspective of the user and the use case. Administrative metadata enables data consumers to access and integrate data into their environment by clarifying data type, format and access rights. Configuration metadata is important for IT to perform data prep or ETL. Application Programming Interface (API) metadata shows a programmer how to integrate data into a website.
  • Integrate and Automate: Interactive metadata supports automated processes for communication and coordination among systems.

Programs and Platforms

Metadata supports reporting and visualization, allowing C-suite members to make better decisions. Metadata enables the transformation of operations allowing the business to grow. Labeling is critical so that data can
move around the organization and be used in innovative ways. Once data is understandable, he said, “You’re going to have people using that data to derive insights that they didn’t even know they didn’t know.”

Data Repositories

Existing information architecture enables – or disables – the depth, scope and quality of available metadata. Adams said that the discussion about repositories is more enterprise architecture-driven, rather than about user needs and business priorities. “It defines what you can do going forward, and it also defines what you cannot do today.”

Reference Architecture

When documenting the Information Architecture, Adams suggests focusing on how the information flows around the architecture of the organization, rather than focusing on specific systems. Start with the type of information and where it resides and denote broad applications and system boundaries. Include data shared with people outside the organization. Although it’s critical to understand what’s happening inside the organization, from a risk perspective, when it comes to risk, it’s more important to understand what’s happening outside the organization. “The interesting thing about this is that you want to use it as a communication tool,” he said. If initially it’s too complex for business users to understand, simplify it a bit. The important thing is to bring people on board.

Data Governance

Often overlooked, governance metadata is Business Intelligence (BI) for your data: metadata about metadata. Metadata ties together Business Strategy, Data Strategy, Data Management and operations with Data Governance. “‘What is the state of my metadata across my ecosystem?’ That’s a bit of a wacky concept for people to grasp.” Enterprise architectures and data reference models are an attempt to align and understand governance policies down to the lower level, Adams said.

Metadata can provide answers to governance questions, such as:

  • How do I know I’m doing this correctly?
  • What constitutes ‘good’?
  • Are we deploying best practices? Are they
    defined?
  • Is this data sufficiently labeled to be
    considered ‘governed data?’

Building Capability

As competitive factors in the marketplace continue to evolve and change, the ability to quickly rise to meet those challenges can mean the difference between success and failure. Developing new capabilities, scaling to
meet demand, and controlling risk requires the ability to pull reports using data in ways that are impossible to anticipate in advance, Adams said. “If that’s the environment you want, then you want well-labeled data that allows you to pivot, schema-on-demand kind of activity, and a very flexible perspective.”

Managing Metadata: An Examination of Successful Approaches

Tags: ,

Automating Data Management and Governance through Machine Learning

21 Aug

MDM Joke

I fairly consistently get questions on the value of Machine Learning in data management and governance. Sometimes this question is framed at a high level in a very “buzz wordy” way. The person asking the question may not know what machine learning (ML) is. They have just heard the words so many times that they know it is good and should be part of the discussion. At other times, the person asking the question knows about ML and various other analytical techniques, but has never really thought of ML in the context of a data management tool. The challenge is that the emergence of IoT data, Customer 360 programs, and emerging best practices that focus on sharing semantically tagged data, all contribute to a fundamental need to do things differently. Machine learning is one of the tools in the toolbox to address the challenges related to scale, change velocity, and the consistent evolution of users and their use cases.

This post focuses on how we can automate the process of identifying data, classifying it, and linking it to internal and external references to provide semantic meaning. The goal of this post is simply to describe what machine learning is for the data manager, and what tasks it performs in the context of the standards based operational perspective.

From an operational perspective, the figure below presents the evolution of data from the “raw” transactional state to a highly labelled or curated state that can be shared between purchaser and vendor; or indeed any producer or consumer of data. Machine Learning plays a role in automating how data is curated and enriched across this lifecycle.

Curation Flow

 

Figure 1: The Curation from raw data to sharable Information

If we drill down on the curation lifecycle, we can identify the various repositories that would be required, and a few of the key supporting standards. These standards and their roles are discussed more completely in a follow on post.

Curation repositories

Figure 2: The Curation flow across the various repositories required. NOTE that this functional perspective is discussed in the context of a metadata hub in Enterprise Data Management – Where to Start?

The database symbols outlined in blue (solid lines) represent data at rest.  The rectangular items outlined in green (dashed lines) represent tasks that automate how data is augmented as it moves along this path. The focus of this discussion is on these green boxes.

Activities within the Data Quality Rules and MDM Rules tasks can be broken down into a number of functional capabilities as detailed below. Some of these capabilities are traditional data operations tasks; namely, persisting metadata in a database, and exposing the data through some sort of cataloging and publishing capability. The other items (outlined in blue) are those where machine learning approaches can be applied.

MDM ML functions

Figure 3: Functional capabilities supported by Machine Learning

First let’s start with a definition of Machine Learning as  Machine Learning has multiple definitions within the popular literature. The website Techemergence provides a comprehensive definition:

“Machine Learning is the science of getting computers to learn and act like humans do, and improve their learning over time in autonomous fashion, by feeding them data and information in the form of observations and real-world interactions.”

Machine learning techniques play a major role in automating the process detailed above especially over unknown or new data sets.

For data management practitioners it is important to understand that no one machine learning technique is going to apply. In all likelihood multiple approaches will be chained together and invariably executed recursively to ensure that the data can be identified, classified and then linked to the appropriate unique identifier. In the ideal world, the algorithms will change or learn to accommodate changes in the data being classified. The figure below lists some of the machine learning techniques that may be applied.

Machine Learning Techniques
Unstructured Data Structured Data
·       Entity Tagging / Extraction

·       Categorize

·       Cluster

·       Summarize

·       Tag

·       Linking

 ·       Associate

·       Characterize

·       Classify

·       Predict

·       Cluster

·       Pattern Discovery

·       Exception Analysis

Note that these invariably interact with one another. If I tag people entities within unstructured text, I may wish to characterize them using structured technique: count of male names; frequency per document; frequency across documents, etc. This speaks to the layered and recursive nature of machine learning, and the richness of the metadata that the data team will need to manage. For a more technical view of ML techniques see this summary.

These are detailed below with considerations for program managers.

Capability Considerations
Identify Machine Learning approaches support the identification of instance data in order to classify the data. Is this personal Information? Does it look like a financial #? Does it reside in a financial statement?

For organizations where there is a significant installed legacy challenge. It will be important to have algorithms that identify data of interest. The identification of personal information is a current area of interest driven by the GDPR regulation.
Classify Once data is identified, ML approaches support classifying the data within the data dictionary: data is in finance domain; it is in the “Deliver” phase of the Supply Chain Operations Reference (SCOR) lifecycle; etc.

Classification algorithms must exist that tag the data with the appropriate classifier. Capabilities must quantify and resolve those instances where there is uncertainty as to accuracy of the classification algorithm. For example, are we are 100% certain that this is a vendor and not a customer?
Resolve The completed data dictionary will support entity resolution by providing a richer feature set against which MDM machine learning algorithms can be run.

Resolving the identity of the master data element may require a multi-tiered approach be run iteratively: apply Algorithm #1; for those that do not resolve with Algorithm #1, apply Algorithm #2; etc.

For example, now that I know that have classified the data item as vendor master data (previous step), can I resolve the identity with certainty to identify which vendor it is?
Link The resolved entity must be linked to internal and external reference sources. Machine Learning techniques may be used to identify and resolve link candidates and specify link type / strength.

The analytical details of this may be addressed in the above “Resolve” capability. However, the focus here should be on identifying the correct link (or links) where there are multiple candidate reference sets where links could be established.

This is a critical step as the linkage to the internal reference “Concept System” is what describes the data element from a semantic perspective. It is also what links the data being described to a publicly available set of definitions that external parties can reference (See “Sharable Information” in figure above). These linkages cross walk an industry accepted definition between supply chain partners.

Example:

If a supply chain manager seeks to communicate the nature of a product requirement to a vendor – a machine screw for example. The ability to specify length of screw versus length of the “shoulder” on the screw; thread size (Metric, standard, imperial?); type of head (hex, square, pan head, etc.) is critical. The internal labels for these are linked to the industry agreed on labels available to the vendor community.

As long as the vendor is using the same reference concept system, both buyer and vendor can be assured that they are talking about the same machine screw.

Once these activities have been completed, the results need to be persisted in a metadata repository and published in a Data Catalog that will allow users to understand what data is available and how it can be accessed.

Some Closing Thoughts :: It’s all about the Ecosystem Maturity!

The above discussion and the content of the two posts in the works on MDM standards and data quality, identifies a set of standards and techniques that seek to streamline and automate the process of Master Data Management. However, these exist within the context of the organization’s data ecosystem. Data practitioners seeking to evolve master data management must ask some core questions regarding information architecture and data management maturity within their ecosystem:

  • How do these standards support my data strategy?
    • Do I have a business case?
    • Executive sponsorship?
    • Funding?
  • Does my information architecture support the capabilities that I need to manage Master Data as envisioned by the standards?
    • Will legacy systems impact how this gets executed?
    • Does the architecture support a “Service Oriented” metadata registry or catalog concept?
    • Do I have a metadata catalog?
    • What are the architectural boundaries and how do I share data across those boundaries?
  • Do I have the data management maturity to execute?
    • Identified and scalable processes?
    • Processes applied consistently across business units?
    • A governance operating model that can accommodate new functions and the change management overhead?
    • What controls and metrics exist? Need to be created?

Understanding how standards and machine learning fit within the information architecture and the organization’s capability maturity will enable the data team to define the right strategy and build out a realistic roadmap. For organizations with an established and mature governance function, many of the above questions will be resolved – or the mechanism to resolve them exists. However, for organizations that have less capability maturity, the strategy and roadmap will need to be explicit in identifying the business units where foundational capabilities can be created that can later be adopted across the organizations as the need and maturity evolve.

For an additional perspective on the business cases that might drive ML, see Classification – the key to releasing data’s value !

Tags: , , , ,

What is Machine Learning?

11 Jul

I enjoyed this post in Techemergence. I was originally looking for a definition of machine learning that helped reconcile all of the different definitions that are out there. I like the approach that they took. I  modified a table they had (presented below) slightly to capture some of the thoughts on machine learning methods.

This is too complicated for many, but captures the idea that ML is layered, and will involve many techniques. I have a simplified list in Automating Data Management and Governance through Machine Learning.

See also No, Machine Learning is not just glorified Statistics for some more discussion in plain English on Machine learning.

Classification Scoring Recommendation / Prediction
·       K-Nearest Neighbor

·       Support Vector Machines

·       Naïve Bayes

·       Logistic Regression

·       Decision Trees

·       Sets of Rules

·       Propositional Rules

·       Logic Rules

·       Neural Networks

·       Bayesian Networks

·       Conditional Random Fields

 ·       Accuracy / Error Rate

·       Precision & Recall

·       Squared Error

·       Likelihood

·       Posterior Probability

·       Information Gain

·       K-L Divergence

·       Cost / Utility

·       Margin

 Combinatorial Optimization

·       Greedy Search

·       Beam Search

·       Branch & Bound

Continuous Optimization

·       Gradient Descent

·       Conjugant Gradient

·       Quasi Newton Method

·       Linear Programming

·       Non-Linear (Quadratic) Programming
Credit: Dr. Pedro Domingo, University of Washington (Slightly Simplified)

Tags: ,

Classification – the key to releasing data’s value !

6 Jun

Someone asked me the other day what the business case was for classifying data. For anyone that has engaged with data to perform analytics or produce business intelligence reports, this may seem like a silly question. However, in many minds, the data does not need to be labelled or classified in any way. The data is used by an application and if that application is performing correctly, the data must be good. And, at some level they are right – as long as the data involved never has to be used outside its application, it may never need to be classified or labelled in any way. The data receives all of it semantic context from the application where it is used.

So when does classification become important? It becomes important when data leaves the application that gave it context. For many of our customers this occurs when data leaves the transactional ERP type system, and is moved into a data warehouse or a data lake whose purpose is to provide access to data from multiple sources. Traditionally, this movement from transactional to a more generally accessible repository came with a level of curation. Prior to the concept of the “Data Lake,” data was moved into the data warehouse with the goal of making it the “single source” of truth. This often involved significant levels of data stewardship and curation to reconcile conflicting versions of “truth.” With the growing awareness and adoption of analytics, the idea of a stable concept of “truth” is elusive. The right data for an analyst is context driven and at times highly variable. The Data Lake construct addresses this issue by allowing all data to be loaded so that the user can determine what data to use based on the decision context at the time. This is what data classification enables. Well classified data can be discovered, analyzed, accessed and integrated into a user’s context based on the classification labels that have been exposed to the user in the Data Asset catalog. Based on this perspective, classification is foundational for driving value out of data in the areas of analytics, business intelligence, operational efficiencies, and compliance.

Indeed in the big data space, classification is foundational for analytics, machine learning, the application of higher level logic, and (way up the maturity curve) for building artificial intelligence capabilities. As a foundational building block for Ai, classification is an interesting topic; although for many too abstracted from today’s problems. However, as the foundation for making data discoverable, understandable, accessible and able to be integrated into downstream applications, it is highly relevant to today’s challenges – almost regardless of where your current capabilities stand. For this reason any data management shop should include in its planning a workstream that seeks to evolve classification capabilities

Consider the following uses cases:

Business Intelligence: marketers seeking to report on price sensitivity and are comparing the difference between prices quoted, prices invoiced, and prices paid net of discount. Data across all of the ERP or transactional systems in use must be classified such that the BI Team is assured that all fields marked as “Price” are the correct type of price.

Marketing Analytics: Your customer 360ᵒ program seeks to understand external factors that may have influenced pricing and discounts provided. What customers are related to the prices referenced above? What kind of customers are they (industry, buying frequency, average purchase, …)? How can I correlate those with external events (elections, new regulation, natural disasters, …)? All of this analysis is supported by data that is classified to reflect the types of queries that may occur and analytical operations to be performed.

Operational Efficiency: Your COO wants to ensure that the acquisition process is fully optimized, and seeks to benchmark operations using the SCOR (Supply Chain Operations Reference) Model. The Operations Team downloads the 250 SCOR performance metrics and seeks to map those to the relevant data. Classification supports the ability to find the right data and map it to the data specified in the SCOR Model.

Compliance & Risk Management. Risk teams will rely on well classified data to enable risk models that are robust and flexible in their ability to address evolving risk. This is especially the case for risk associated with adaptive threats; for example fraud and cyber-crime.

Bottom line, if classification is not something that you have thought about, consider putting a plan together. It is the key to releasing the value of your data, and fully leveraging data as an asset.

Metadata

 

 

Tags:

Building Solid Foundations in Big Data & Analytics

23 Aug

Originally Published on the DATUM, LLC Site: Building Solid Foundations in a data Swamp

Much has been written about Big Data, Data Science and Artificial Intelligence and how these will change the world through the insights being derived from the data. This especially applies to the unstructured data. A recent article in the Harvard Business Review indicated that “cross-industry studies show that on average, less than half of an organization’s structured data is actively used in making decisions—and less than 1% of its unstructured data is analyzed or used at all.”[1]

There are a few challenges however:

  1. How do users create understanding and ensure they have the correct data for their needs if it has no structure?
  2. How do you create a single logical view of data in a big data world, where things are not only highly variable, but also are often widely disbursed.
  3. How do you address analytical requirements, where the notion of data quality and how it is managed, varies significantly?
  4. How do you expose the data lake(s) to users in a form that is discoverable, understandable and useable?

This blog is the first in a series to explore the data management and governance perspectives related to these four challenges.

Challenge #1: Unstructured Data

The question of how to deal with unstructured data consistently raises its head as a challenge for organizations. First let’s get a few things out there:

  • There is no such thing as truly unstructured data. There is always a structure of some sort.
  • Knowing what you have and having the right tools are foundational capabilities.
  • The degree of structure required for data to be useful is variable and context driven.

Let’s take these in order:

Creating Structure

Structure is created in one of two ways:

  • Through reorganizing data so that it has structure
  • Through labeling data

The former is what happens to data in a traditional data environment as it is moved through the ecosystem – from Source to Enterprise Data Warehouse for example. The latter is what happens in a big data environment. The data is never moved, but rather labels are added to it to provide the ability analyze that data.

Note: Data can be labeled incrementally. Newly acquired data, can only be labelled with the acquisition date, the source, and the file type. As data moves through the data lifecycle, it will be “curated” to add additional context.

A little labelling goes along way!

How much the data needs to be labelled to be useful can be viewed on a continuum. At one end simply knowing that you are looking at emails provides enough information to know how to organize them; while at the other end, social media sentiment analysis will require extensive labelling. Regardless, the right tools are required to provide logical structure to the unstructured data.

When it comes to tools that cater to unstructured data one key capability is entity tagging or entity extraction tools that can recognize an entity and tag it with a label that makes sense to the organization – essentially tag it with the approved glossary term. Entities can be:

  • Anything from a simple named list such as a “product”; or
  • Extremely complex and map entities into semantic ontologies such as a “JV” is a “Joint Venture”, which is a type of “Company”, which is an “organization” that has “owners”.

Complementing the tagging capability is a flexible indexing capability. Tools like Elastic Search allow users to search based on the structures discovered in the data.  For example, a “Joint Venture “is a type of company. Additionally, these tools can create an index to allow discovery of similarities in text.

The key point is that once data is organized, users and applications can begin to apply big data techniques to expose insights:

  • How do emails cluster on a timeline?
  • Are organizations mentioned in the text? (Could be Joint Ventures, Partnerships, LLCs, PLCs, and so on.)
  • Is there a change in frequency over time? Related to what entity types / categories?
What does this mean from a data management perspective?

From a data management perspective unstructured data will require some new capabilities. However, in some respects, it really is more of the same: What data do I have and where is it?  Is my data labelled to communicate understanding?  Is my data easy to acquire and apply in my context?

If you think of tags or labels as descriptive metadata, and the list of tags and labels as reference metadata, then you can place this activity into the traditional data management context. In order for data to be discovered, understood and integrated across systems and use cases, organizations need to:

  • Have a disciplined approach to how data is described and labelled. This starts with creating a set of glossary terms that can be linked to define meaning. [2]
  • Implement the governance framework that ensures the data is aligned to – and remains aligned to – the business understanding of what the data is, and how it is used.

Organizations often do not face this challenge until they need to manage data across the various operational silos, geographic regions or functional domains. The need to understand product lifecycle data with regional focus group data is an example of a cross functional/geography/silo data mash up that delivers high impact insights.

Be sure to check back in as we address the next three challenges!

References

[1] Harvard Business Review What’s Your Data Strategy? Leandro DalleMule, Thomas H. Davenport; May –June 2017 Issue https://hbr.org/2017/05/whats-your-data-strategy

[2] With reference to linking of data, the simple link types are “subset of”, “superset of”, “same as”. (See SKOS for a deeper discussion on knowledge organization). For example, using this approach one can tag pharmaceutical products to identify synonyms as recognized by the ISO standards; and synonyms of the same product that are commercial names. This is the challenge faced by organizations implementing the IDMP standards.

[3] For a good case study of data integration across disparate data sets using SKOS metadata see Healthcare Research Information

Tags: , , , , ,

The evolving buzz phrases: “Integrated Analytics”, “Analytics as a Service”, “Analytical Centers of Excellence”…

2 Aug

OReillyCover“Integrated Analytics”, “Analytics as a Service”, “Analytical Centers of Excellence” these are all buzzwords that while they have been around a while, seem to be gaining traction. I downloaded the eBook Integrated Analytics: Platforms and Principles for Centralizing your Data from O’Reilly publishing. Generally I do not pay much attention to these books as they seem to result in calls from the sales departments of the product companies mentioned in the book. However, I wanted to see what they had to say . First – pretty good for a 17 page book, and probably worth a read over the morning coffee; but a few points.

 

Point #1: This is a paper about analytics, and yet very quickly it becomes a paper about data – data is not sexy, so the paper does not position it itself as a data paper. However, the implementation reality is that analytics especially in the context of the buzzword phrases we are talking about is a data problem – more specifically a data management problem.

Point #2: The paper has a number of good referenceable ROI numbers. Specifically from Nucleus Research. These are always handy for use in presentations, and in supporting the business case process especially if you are setting up an enterprise level analytics or data management function.

Point #3: I was struck by the position in the paper that all data should be centralized. On the one hand there is this recurring theme that the data warehouse is rigid, time consuming, costly, slow and requires specialized skills. The shortcoming of the data warehouse approach are focused on the fact that it is a centralized source of data. On the other hand the paper is recommending that all data be centralized. It does not really address the underlying issue that the recommendation to solve the problem of the centralized data  is to centralize the data – in presumably a new way?

 

When it comes to #3, the paper is not wrong, its just that the reality is that data in large organizations will always be silo’d. The underlying operational / transaction oriented systems that drive business will always be optimized for the operational environment, and rarely for analytics. Indeed structuring the data for analytics in the traditional sense (i.e. dimensional cubes) implies that the analytics has already been done. The true analyst has moved onto the next problem, and is looking for meaning and insights in the data that are not yet known. This article by Tip Clifton talks to this challenge nicely.

What the paper glosses over is that regardless of where the data resides, one needs to know of its existence and nature. What is key is the centralization of the knowledge regarding the data you have – not necessarily the data itself. Indeed fundamental to many of the “big data” discussions is the notion that the data does not move. It is used “in situ” as they say. The knowledge that needs to be centralized centers around the following:

  1. The context it was collected in. Was this data  collected through social media, through a news agency, or from an established research organization?
  2. How it can be used. Population estimates created through sampling performed by the US Census are going to be used in a different context than population estimates created through location tags on twitter feeds for example. One is not better than the other, but you would not use them in the same analytical context.
  3. The level of curation. Did anyone validate this data asset? Does it have a shelf life? Who maintains it?
  4. The  lineage and provenance. Is this a referenceable data asset? will my analysis be transparent, testable, repeatable and defensible? Will the analytical conclusion hold up under scrutiny: in front of a judge; your budget committee; your investors; your customers?
  5. How to access. Where is this data asset located? How can it be accessed? by whom? Under what conditions?

This centralized repository of the above knowledge can come in many forms, but generally speaking we are talking about metadata. The premise here is that if we have fully described and well cataloged data, it does not matter where it resides.  Indeed, the work involved in getting the data centralized and structured  is liable to devolve into the much maligned Data Warehouse ecosystem that the paper is suggesting we avoid.

I discussed this notion of the managed metadata environment of this nature in this post which discusses how one gets started. Inherent in this approach and well addressed in the paper are a number of data management aspects that are critical: the data plan; governance; stewardship; and tools. The notion of the overarching architecture is missing. However, this needs to be considered as the business architecture will drive system architecture and the set of tools selected. Of course the legacy architecture will often play a huge role here.

When I am talking to clients on the topic of building out analytical capabilities, I tend to focus on three things in order of importance: The business case and impact associated with the analytical capabilities being developed; the analytical approach that makes sense for your organization; and the underlying data foundation on which the analytics must reside.

For integrated analytics, and the buzzwords we are discussing, the data foundations are built on the best practices embedded in the CMMI Data Management Framework aligned to the traditional analytical cycle. This alignment of the CMMI framework specifies  the key management practices required to identify,  build, scale and manage the integrated analytical environment; whether it is implemented as a service, in a Center of Excellence, or some combination of the two.

Download the paper and have a read. It provides a good starting point.

Tags: ,

Enterprise Data Management – Where to Start?

3 Jul

What is Enterprise Data Management? In some organizations, this is an easy question to answer. However, in others – especially those with an analytical mission – it is much harder. Often the function is put under the Enterprise Architecture team.  One often hears that “the Enterprise data folks just do not get it”. As one executive in a large financial organization put it: “EA is where rubber hits the air”. So how do we define a role for the data function within an Enterprise Architecture  team?

This post is not about how to organize effectively in order to align with the business units to show business impact – although a worthy topic. This post is about suggesting a role for the enterprise data management team when that team is organized under the CIO within the Enterprise Architecture  function of the organization. In order for the data function to be deemed valuable to the business stakeholders it must be understandable, actionable and  tied to the business objectives.

Data is everywhere. When we talk about “Enterprise Data Management” the temptation is for managers to say – well that means we manage all data in all locations. As enterprise data managers, we must know all about everything! Really? Have you ever seen this work? This leads to the top down mandate of the “canonical” approach where the objective is a single standard, a single canonical model – a single ring to rule them all! This rarely works well (if at all). Business requirements, analytical activity, market trends, and evolutions in technology all lead to a core business requirement for flexibility. Additionally, there is a fundamental need to recognize that the “ground truth” is almost always with the business side of the house – and “truth” is often a shifting concept in the real world. This is part of the reason why the Enterprise Data Warehouse (EDW) “single version of the truth” is problematic for analytical and BI staff and for the rise of Hadoop as a more flexible environment. As an analytical or BI person, my version of the truth – or the right data – depends on the context of a particular decision.

So how do we focus the EDM team on what makes sense? The graphic in this excellent article on risk architecture caught my eye. I have modified it a bit to identify some core activities that I see as foundational for organizations seeking to mature their data management in general, and specifically, the integration of the enterprise architecture team in the data management process. The original graphic is attributed to Naomi Clarke currently at Credit Suisse.

EDM

Based on the above, the role of EDM is simple; manage data assets to expose those attributes that are needed to answer key business questions about data assets:

  • What are they?
  • Where are they?
  • What has happened to them?
  • What are they related to?

To do this, one needs a data management “hub”. I call it a Hub as this provides flexibility for discussion purposes. Some would call it a Managed Metadata Environment (MME), others perhaps a Metadata Registry. Regardless, the goal is the metadata ecosystem that can support key functions related to governance, curation, quality, usability and discoverability. This view suggests the following regarding the roles of the EDM team – especially when it is organized within the EA Team:

  1. The Team needs to only manage three inputs: lineage metadata; definitions, and the physical location (what, where and change). The way the organization creates those three inputs is part of an overall data strategy, but not something the EDM team drives – these are driven by the business. By focusing in this way, the Enterprise team leaves it up to the business or operational components to determine the optimal approaches.
  2. Definitions are aligned to business terms and to “Concept Systems” (as defined in the ISO 11179 Specs). This enables discoverability and complex search approaches based on an understanding of semantic equivalence.
  3. Data Assets can be classified within the context of an enterprise  data reference model (DRM). In most organizations, this supports the governance process. However, in government organizations, the DRM is also used to align policy and strategy objectives to IT activities. See Federal Enterprise Architecture framework for how this works in the US.
  4. Capabilities to support governance functions must be provided: vocabulary management tools with the capability to curate and link ontologies, taxonomies, controlled vocabularies, etc, data quality tools, and governance tools.

If one can limit the INITIAL scope of the EDM team to these items, it is much easier to tie enterprise activities to the business needs, and provide a set of capabilities that address challenges of high value to the organization: search, discoverability and integration. Evolving the role once these benefits are established is a much easier task.

Tags: , , , , , , , ,

A good baseline presentation on creating semantic data repositories

8 Jul

I have been meaning to give a shout out to Dan MCCreary http://www.danmccreary.com/ for some time now, and am only just getting around to it. What caught my attention was a presentation Dan gave at the 2006 Semantic Technology Conference. The presentation walked the audience through the process of creating a well-structured data repository enabling players within the Minnesota education system to ask complex across all of the data available. While the presentation is rather dated, it is one of the better presentations I have seen that takes things back to a level most people can understand. Additionally, I don’t see that the state of the industry has changed that much in terms of tools to support the creation well-structured data repositories that support broad “knowledge management” goals.

Dan focuses on metadata as the means to achieve the goals of the participants: Minnesota Department of Education; the Wisconsin Department of Public Instruction and the Michigan Department of Education. The goal is to create a semantic understanding of the data available. Semantic is one of those words that means many things to different people. Once you combine “semantic” with “knowledge management” or ”knowledge discovery”, everyone seems to have a different opinion as to what you are talking about – IT and business folks alike. It becomes very important to level set your audience. For the purposes of this discussion, creating a semantic repository involves:

  • Enhancing data with metadata that focuses on describing the data asset as a whole, and with respect to its component content parts;
  • Creating descriptive and content metadata that describes the data in specific terms, and in more general “conceptual” terms.
  • Creating a description of the data that reveals connections to other concepts; related terms; multiple standards.
  • Creating highly linked data that is organized through taxonomies and ontologies to reveal links that are perhaps not obvious to users.

The desired end state is the ability for a broad range of end users to query disparate data sources and understand how available date can be used in their particular context.

Within the referenced deck (http://www.danmccreary.com/presentations/semweb2006/), there are a number of sections worth drawing your attention to:

ISO 11179 (Starting on page 15). Most people I speak to that know about ISO 11179 (http://www.iso.org/iso/home/search.htm?qt=11179&published=on&active_tab=standards&sort_by=rel ) are somewhat skeptical as the spec is a heavy on as written. Once you get your head around all the good ideas that have gone onto the specification, the question becomes how much of it do you adopt? The idea behind the spec is that each piece of metadata has its own metadata that allows a user to know exactly how that metadata is to be used. Dan presents the 11179 concept within his discussion of the National Information Exchange Model (NIEMS https://www.niem.gov/Pages/default.aspx) which brings us to another interesting point…

Mapping to Standards (Page 16). I like the way that the process of mapping to standards is presented. In this instance, the Educational System is only trying to map to one standard. However, you can imagine an environment where there are multiple standards of interest. In this instance, the mapping approach is organized around the OWL “SameAs” terminology (Web Ontology Language http://www.w3.org/TR/owl-features/). It does not say it in the presentation, but I am assuming that this is taken from the SKOS Simple Knowledge Organization System standards (http://www.w3.org/2004/02/skos/). As you classify data assets, there will be multiple mapping exercises where the mapping will not be exact. The linking concepts defined within SKOS will allow you to relate items that are not precise or consistent in their relationships.

Demonstrating that the data problem grows faster than the number of data sources (Page 17). This sounds so simple. However, I am still surprised when I hear people say – “we have the data mapped between the sources I want, so no need to spend time mapping to standards”. Well ok that works if you have a small number of static data sets, but … that happens rarely. In the presentation Dan refers to this as the O(N2) problem versus the O(N) problem. I have included his graphics as these tell the story better than words.

network Problem

Tags: , , , , ,

Me, My Data & I

29 Jan

me at the center

One of the ways in which big data is moving into the mainstream is through a growing culture of self-measurement. We see this through the smartphone apps that track the number of daily-diet calories added by lunch at Chipotle. New fitness watches notify us when we sit idle for too long, and track the amount of light vs. deep REM sleep we get each night. When a sudden, out of the ordinary, credit card purchase sparks a stolen-card call from the bank, we are reminded that day-by-day, quietly and invisibly, our individual level data is collected, stored, and even analyzed without our conscience awareness.  A recent article in The Guardian makes interesting commentary about this phenomenon. It focuses mainly on healthcare, but I think it has larger implications for individual-level data outside of the sector.

You can find the article here: http://www.theguardian.com/science/political-science/2014/jan/27/science-policy

In a nutshell, the author writes about an experience with error in her smartphone’s running app, which prompts her to question and seek to review the previous data collected about her through the application. Of course, like most apps, this one does not allow her to retrieve the data collected about her by the system.  Instead, she and other users will experience the cumulative effect of data collected through new products and updated apps. After several examples that illustrate the potential value of allowing users to access to data collected through such applications, the author concludes that doing so could improve the health of everyone in addition to generating better products.

I think the article raises a valid question about big data that few people discuss. Don’t we miss opportunities for value creation by not allowing individuals to access the data collected about them? A very obvious example in healthcare would be the case of individuals living with type II diabetes – a disease whereby the body’s insulin no longer effectively brings glucose into its cells. Because of this condition, individuals living with type II diabetes  depend dramatically on lifestyle and other daily choices for their health. Liberating individual-level data collected through diabetes apps might prove particularly beneficial, both for individuals and for the health system, if people are allowed to use the data to find their own creative ways of improving their health.

However interesting this question may be, I’d like to take it a step further. What potential dark clouds might arise in an otherwise sunny horizon for big data if people aren’t allowed to access data collected about them? One potential problem is that the players who have authority to access and analyze this data may use it to pursue objectives that create value only for themselves. In healthcare, this might manifest as a company using its big data insights to aggressively market to a particular demographic that predominantly suffers from a given disease. What is the downside in this scenario? For starters, many of the recipients of this aggressive marketing may resultantly seek and receive unnecessary treatment that may prove harmful to their health, while those in other markets who actually need the treatment may go without it due to unequal exposure. Another potential downside stemming from this scenario is that it potentially destroys value, especially if payers spend more money on treatments that do not generate value for patients.

In the end, it is good food for thought. Beyond reaping the fruits of product development, consumers might stand to benefit from having access to big data that they help to supply.

Tags: , , , , , , , , ,

Metadata related Links & Standards

15 Jan

This is a work in progress, and I will keep updating it, but it is a place to start…

Name Comments
MARC 21

  • Classic MARC (MARC 2709)
  • MARC XML
  • A data element set defined by content designation and semantics
  • Institutions do not store “MARC 21”, as it is a communications format
MODS:Metadata Object Description schema
  • An XML descriptive metadata standard
  • A derivative of MARC

ü  Uses language based tags

ü  Contains a subset of MARC data elements

ü  Rich, but not as rich as full MARC

  • Does not assume the use of any specific rules for description
  • Slowly gaining acceptance as an alternative to MARC in digital library environments
Dublin Core
  • Maintained by the Dublin Core Metadata Initiative
  • DC Element Set (or Simple DC) – 15 elements, all optional and repeatable
  • DC Metadata Terms or Qualified DC – Simple DC + additional elements, refinements, and encoding schemes
VRA Core
  • Maintained by the Visual Resources Association Data Standards Committee
  • Meant to describe both the original piece and the visual representation (photograph, slide, digital resource)
  • Content guidelines: Cataloging Cultural Objects (CCO) – now in draft form
IEEELearning Object Metadata
  • Meant to describe ‘learning objects’, i.e. “small units of instructional content that can be reused in different contexts.”
TEI Header
  • Part of the Text Encoding Initiative Standard (an XML based mark-up language for texts)
  • Meant to provide the descriptive information about the text itself, its source, its encoding, and its revisions.
  • More than just descriptive metadata: technical and administrative
EADEncoded Archival Description
  • Meant to mark-up archival finding aids
  • Hierarchical collection description
Dublin CoreCollection Description Application Profile
  • Unitary collection description
  • No standard scheme yet, but Dublin Core Collection Description Application Profile is close
  • Growing importance in federated and aggregated search environments
GILSGovernment Information Locator Service
ONIXOnline Information eXchange
  • publisher driven standard for exchanging information in electronic form
FGDC CSDGM Federal Geographic Data Committee ContentStandard for Digital Geospatial Metadata
  • Describes digital spatial data
DDIData Documentation Initiative
  • Social science datasets
METS
  • The METS schema is a standard for encoding descriptive, administrative, and structural metadata regarding objects within a digital library, expressed using the XML schema language of the World Wide Web Consortium
MPEG21
  • Industry standard; operates similarly to METS. Meant for multimedia applications; includes rights and IP information.
CWM
ISO 3166
Premis
  • The PREMIS Data Dictionary for Preservation Metadata is the international standard for metadata to support the preservation of digital objects and ensure their long-term usabaility.
  • http://www.loc.gov/standards/premis/
SDMX
SKOS
  • SKOS is an area of work developing specifications and standards to support the use of knowledge organization systems (KOS) such as thesauri, classification schemes, subject heading systems and taxonomies within the framework of the Semantic Web.
  • Based on RDF – Resource Description Framework
  • http://www.w3.org/2004/02/skos/
FIBO
ISO/IEC 11179
  • ISO/IEC 11179, Information Technology — Metadata registries (MDR)
  • Part 1: Framework
  • Part 2: Classification
  • Part 3: Registry metamodel and basic attributes
  • Part 4: Formulation of data definition
  • Part 5: Naming and identification principles
  • Part 6: Registration
ISO/IEC 19763
ISO 25964-1:2011
  • ISO 25964-1:2011 also provides a data model and recommended format for the import and export of thesaurus data.
  • ISO 25964-1:2011 is applicable to monolingual and multilingual thesauri.
  • ISO 25964-1:2011 is not applicable to the preparation of back-of-the-book indexes, although many of its recommendations could be useful for that purpose.
ISO 4217
  • ISO 4217:2008 specifies the structure for a three-letter alphabetic code and an equivalent three-digit numeric code for the representation of currencies and funds. For those currencies having minor units, it also shows the decimal relationship between such units and the currency itself.
  • ISO 4217:2008 also establishes procedures for a Maintenance Agency, and specifies the method of application for codes.
  • ISO 4217:2008 is intended for use in any application of trade, commerce and banking, where currencies and, where appropriate, funds are required to be described. It is designed to be equally suitable for manual users and for those employing automated systems.
MOFISO/IEC DIS 19502
  • The Meta-Object Facility (MOF) is the OMG’s adopted technology for defining metadata and representing it as CORBA objects using UML.
  • Do we care about how things get presented in CORBA? It is about how software works together
International Classification of Diseases (ICD)
XBRL
IFRS
  • The IFRS Taxonomy is the XBRL representation of IFRSs, including International Accounting Standards (IASs), Interpretations, and the IFRS for Small and Medium-sized Entities (SMEs), issued by the IASB. The IFRS Taxonomy contains tags for all IFRS disclosures.
  • http://www.ifrs.org/XBRL/IFRS-Taxonomy/Pages/IFRS-Taxonomy.aspx

Tags: , ,

← Older Entries