CO1 – Session 5

Session Topic: Design Principles for Cognitive Systems

COGNITIVE COMPUTING
(Course code: 18CS3272)
Design Principles for
Cognitive Systems
• In a cognitive computing system, the model refers to the corpus and
the set of assumptions and algorithms that generate and score
hypotheses to answer questions, solve problems, or discover new
insight.
• The corpus is the body of knowledge that machine learning
algorithms use to continuously update that model based on its
experience, which may include user feedback.
Steps involved in designing the system
• It requires an understanding of the available data
• The types of questions that need to be asked
• The creation of a corpus comprehensive enough to support the
generation of hypotheses about the domain based on observed facts
Why system is designed
• A cognitive system is designed to
• Create hypotheses from data
• Analyze alternative hypotheses
• Determine the availability of supporting evidence to solve problems.
• By leveraging machine learning algorithms, question analysis, and
advanced analytics on relevant data, which may be structured or
unstructured, a cognitive system can provide end users with a powerful
approach to learning and decision making
• The design of a cognitive system needs to support the following differentiating
characteristics:
• Access, manage, and analyze data in context.
• Generate and score multiple hypotheses based on the system’s accumulated
knowledge. A cognitive system may generate multiple possible solutions to
every problem it solves and deliver answers and insights with associated
confidence levels.
• The system continuously updates the model based on user interactions and
new data. A cognitive system gets smarter over time in an automated way.
Components of a Cognitive System

Architecture of a cognitive system

 Building the Corpus
• A corpus is a machine-readable representation of the complete record of a particular
domain or topic.
• In a cognitive computing application, the corpus or corpora represent the body of
knowledge the system can use to answer questions, discover new patterns or
relationships, and deliver new insights.
• The contents of this base corpus constrain the types of problems that can be solved, and
the organization of data within the corpus has a significant impact on the efficiency of the
system.
• you need a good understanding of the domain area for your cognitive system before
determining the required data sources.
Questions to be Addressed in design phase of system
• Which internal and external data sources are needed for the specific domain
areas and problems to be solved? Will external data sources be ingested in
whole or in part?
• How can you optimize the organization of data for efficient search and analysis?
• How can you integrate data across multiple corpora?
• How can you ensure that the corpus is expanded to fill in knowledge gaps in
your base corpus? How can you determine which data sources need to be
updated and at what frequency?
• The choice of which sources to include in the initial corpus is critical.
• Sources ranging from medical journals to Wikipedia may now be
efficiently imported in preparation for the launch of a cognitive system.
• In addition, it may be equally important to ingest information from
videos, images, voice, and sensors. These sources are ingested at the
data access layer
• Other data sources may also include subject specific structured
databases, ontologies, taxonomies, and catalogs.
• If the cognitive computing application requires access to highly
structured data created by or stored in other systems such as public or
proprietary databases, another design consideration is how much of
that data to import initially.
• It is also important to determine whether to update or refresh the data
periodically, continuously, or in response to a request from the system
when it recognizes that more data can help it provide better answers
• In many fields, taxonomies are used to capture hierarchical relationships
between elements of interest
• An ontology is similar to a taxonomy, but generally represents more
complex relationships, such as the mapping between symptoms and
diagnostic criteria in the Diagnostic and Statistical Manual of the American
Psychiatric Association.
• When such a generally accepted taxonomy or ontology exists in a field, it
may be useful to import that structure with its data—in whole or in part —
rather than creating a novel structure for the same data
• The choice of data structures can greatly impact the performance of the
system on repetitive tasks such as knowledge retrieval for generating and
scoring hypotheses
• A data catalog, which includes metadata such as semantic information or
pointers, may be used to manage the underlying data more efficiently.
The catalog is, as an abstraction, more compact and generally faster to
manipulate than the much larger database it represents.
Corpus Management Regulatory and Security Considerations

• Data sources and the movement of that data are increasingly

becoming heavily regulated, particularly for personally identifiable
information.
• Cognitive computing applications learn and derive new data or
knowledge that may also be subject to a growing body of state,
federal, and international legislation.
• When the initial corpus is developed, it is likely that a lot of data will
be imported using extract-transform-load (ETL) tools. These tools may
have risk management, security, and regulatory features to help the
user guard against data misuse or provide guidance when sources are
known to contain sensitive data
Bringing Data into the Cognitive System
• Unlike many traditional systems, the data that is ingested into the corpus is not
static.
• Build a base of knowledge that adequately defines your domain space.
• As the knowledge base increases the importance of the data should also
increase.
• Converting the model to cognitive system implies refine the corpus.
• Therefore, you will continuously add to the data sources, transform those data
sources, and refine and cleanse those sources based on the model development
and continuous learning.
Leveraging Internal and External Data Sources
• Transactional systems and business applications increases the structured data
and text contained in forms or notes and possibly images from documents or
corporate video sources increases the unstructured data.
• A cognitive computing system generally needs to access a variety of frequently
updated sources to keep current about the domain in which it operates.
• Professionals learning from news data
• A popular magazine with articles on psychology
• Healthcare
• Telecommunications
Data Access and Feature Extraction Services
• Any data to be imported from external sources must come through
processes within this layer (refer diagram).
• Cognitive computing applications may leverage external data sources in
formats as varied as natural language text, video images, audio files,
sensor data, and highly structured data formatted for machine
processing.
• The feature extraction layer has to complete two tasks.
• First, it has to identify relevant data that needs to be analyzed.
• The second task is to abstract data as required to support machine learning.
• Any data that is considered unstructured—from video and images to
natural language text—must be processed in this layer to find the
underlying structure.
• Feature extraction and deep learning refer to a collection of techniques—
primarily statistical algorithms—used to transform data into
representations that capture the essential properties in a more abstract
form that can be processed by a machine learning algorithm
• These layers appear as a straightforward process of importing and
refining data, it should be noted that external sources may be added or
removed based on their value in hypothesis generation and scoring over
time.
• Eg. Medical Diagnosis system may add a new external source of case files
or delete a journal if it is found to provide unreliable evidence
• For cognitive systems that provide evidence to support hypotheses in
regulated industries, the data access layer processes or the corpora
management services should maintain a log or other state data so that
an auditor can determine what was “known” at any point in time.
• This would be important, for example, if a recommendation from the
cognitive system was followed by a practitioner and resulted in harm
Analytics Services
• Analytics refers to a collection of techniques used to find and report on essential
characteristics or relationships within a data set.
• The use of an analytic technique provides insights about the data to guide some
action or decision.
• A number of packaged algorithms such regression analysis are widely used within
solutions.
• Within a cognitive system, a wide range of standard analytics components are
available for descriptive, predictive, and prescriptive tasks within statistical
software packages or in commercial component libraries.