Knowledge as a service

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

Knowledge as a service (KaaS) is a computing service that delivers knowledge to users. In relation to KaaS, knowledge is data + context, that is, data with the addition of information that enables users' understanding of the data - definitions and semantics - its relation to the users' situation or the situation in which the data was produced or that which it aims to provide information about.

Knowledge as a service is one of the series of "... as a Service" computing services.

Overview[edit]

KaaS is a new type of "...as a Service" offerings that has been discussed with only nascent examples demonstrated in recent computer (2019) science conferences, in particular ISWC '19, the 18th International Semantic Web Conference.[1] At that conference, it was described how knowledge can be made live and evolve on the web allowing users to learn directly from elaborated knowledge, now appearing in the form of knowledge graphs (KGs).[2] KaaS appear when KGs are accessed via services This is opposed to DaaS which might "compute large volumes of data; integrate and analyzes that data; and publish it in real-time, using Web service APIs" (from Data as a Service) where the KaaS is able to exploit context - both the context of the user in relation to their information requests of the KaaS (where and when they make the request) and also the context of the information in relation to some objective or purpose of the users either understood by the KaaS automatically or indicated to it by the user.

KaaS is described as being more related to Data as a Service, Content as a Service and other services which supply information to users, more than other *aaS, such as Software as a Service which provide functionality however, the idea that a KaaS may analyse context at query time indicates that there is overlap between KaaS and *aaS such as Search as a Service and that perhaps not all KaaS responses are idempotent since their results depend on a context that might be time-dependent.

Differentiating knowledge from data[edit]

As per the description in the overview above, KaaS is differentiated from DaaS or other information *aaSes by its delivery of "knowledge" as opposed to "data" where knowledge is defined by relations of data/information to context that is relevant to the user. This then indicates that the sensible description of KaaS as a service type is dependent on the differentiation of knowledge from data and/or information.

Conceptual models that make such a differentiation such as the so-called DIKW pyramid have existed for perhaps more than 40 years (see a 1974 journal article about this[3]) however definitions are not stable and universally accepted (see the discussion about the conceptualizations of DIKW within the DIKW Wikipedia article that question value of wisdom). The knowledge component of DIKW is generally agreed to be an elusive concept which is difficult to define, however Rowley 2007, in a well known student textbook[4] differentiated knowledge from data by stating that knowledge is "defined with reference to information" and that it contains more than just facts but also "beliefs and expectations".

In relation to knowledge graphs, knowledge may be additional content they provide over and above pure data which is the definition of the categories, properties and relations between the concepts, data and entities that substantiate one, many or all domains of discourse (see the definition of Ontology).

The ability to represent "beliefs and expectations" or other forms of not so straightforwardly explicit knowledge is an on-going area of improvement in information sciences (see Tacit knowledge) and, with relation to KaaS, the establishment of recent informatics mechanics to do so it critical to the legitimacy of KaaS as it is differentiated from just value-added DaaS.

Knowledge graphs' ability to represent context via the definition of the categories, properties and relations between the concepts, data and entities that substantiate one, many or all domains of discourse that they provide (see the definition of Ontology) has led to the idea that supplying access to KNs might be a required competency of a KaaS.

Delivery of knowledge[edit]

Much service-delivered content is dependent on a session to provide much of the context that the user (client) needs to understand answers to questions. For example, using current HTTP internet protocols, a GET request to retrieve information identified by a URI, such as a web page, a client (a human or a machine) may have access information supplied automatically to enable that client to bypass paywalls or other content access controls. Such context, in this case about the client's information access allowances, can alter the information provided.

In a logical extension to this internet protocols example, a server would receive from the client, either manually or automatically, a full context which would be information about the situation the client is in and this would allow the server to best interpret the client's request. Current internet protocols allow for formats, languages and related preferences to be expressed by clients but make no mention of what a client already knows and what they may understand. The recent Content Negotiation by Profile[5] proposes additions to both the HTTP internet protocols and related services that allow clients to also request information - a response from the server - that accords with an identified information model. This then allows clients to indicate not just formats and languages that they understand (technically that they prefer) but also domains of discourse that that do, which is a step towards comprehensive client context provision.

Service mechanics[edit]

With the differentiation of KaaS from DaaS - the data/knowledge difference - the crucial test points used to identify KaaS are not the service mechanics, e.g, whether the service is stateful or uses an API style such as REST but instead that its content is differentiated from that of a DaaS. If this proposition is accepted, it would be possible to find a DaaS and a KaaS that use the same data source and even implement the same mechanics perhaps such as GraphQL, but deliver different results.

This notwithstanding, if a service to be a KaaS must deliver knowledge then the mechanics implemented must not exclude that possibility. This may prevent mechanics that rule out annotations or definitions for packets of data delivered or perhaps data delivered that is not able to be deterministically associated with the question asked since that would destroy result context.

Proposed KaaS Modes[edit]

This section proposes generic modes that KaaSes do or might use to differentiate themselves from DaaS or other services.

Current expressions of KaaS characteristics[edit]

A survey of current (January, 2020) websites and vendors using the phrase KaaS indicate that no single model of KaaS is shared by all those using the phrase. Some aspects of the service suggested as typifying it and differentiating it from DaaS are:

  • "...built on the foundation of decades of experience..." [6]
  • "...putting relevant content at your users’ fingertips." & "Like SaaS, PaaS, and IaaS, Knowledge as a Service is a 'managed service' concept in which a technical service and support organization outsources the development, management, and analysis of knowledge solutions to a third party." [7]
  • "Knowledge as a Service ... accesses ... Knowledge Graphs ... to enhance data and processes through the addition of context." [8]
  • "Knowledge-as-a-Service ... combines people, process and technology to convert, deliver and continuously optimize the knowledge required for [specific situations]" [9]
  • "The customer can get the benefit of professional knowledge without the cost of direct professional service" & "KaaS isn’t a new innovation. Professionals in all types of niches have been peddling knowledge for centuries. But the delivery methods and ways professional service firms consider KaaS are evolving, and that’s due in part to the growth of digital disruptors." [10]
  • "Delivering Knowledge as a Service – KaaS, is not the same as Software as a Service (SaaS) although both share the foundational aspect of being cloud based. KaaS is about delivering the right knowledge to the right person in the right context at the right time..." [11]

Some of the definitional overlaps that can be seen in the quotes above are:

  • data + experience or some other form of knowledge such as professional skill separates a KaaS from just a DaaS
  • context of data or knowledge or of appropriate delivery is a form of knowledge and allows a KaaS to better target results that other services

Systematic listing of potential KaaS modes[edit]

The following image describes Data as a service and processing services, such as the Web Processing Service's components.

Generalised components of a DaaS and a WPS with (left) a DaaS including an access API and a data source, (centre) a Web Processing Service including a processing application and, very likely but not essentially, a data source, (right) the general DaaS/WPS scenario where data/processing is supplied without clients being aware of the system componentry

The following diagram indicates that a KaaS is a DaaS or WPS "+", that is, a DaaS or WPS plus some additional element that realises the "knowledge" offering of the service.

A KaaS is widely regarded to be more than just a DaaS or a WPS. It is thought to contain additional information or calculation, based on expert experience, large datasets, reference data or something other than the primary service data that raises the KaaS offering from merely data/information/processing to knowledge

The following diagrams provide several potential modes of KaaS operation in which different sources of the information/calculation that separates a KaaS from a DaaS or a WPS are utilised. A real-world KaaS might indeed utilise multiple.

Different modes of a Knowledge as a Service that all imbue its outputs with something beyond the data/information/computation offerings of a Data as a Service or a Web Processing Service. (left) Experience (of type unspecified) is added to a DaaS or a WPS, (centre) Additional client data is sourced by the KaaS to better understand the client's context, (right) Reference Data is accessed by the KaaS which allows it to provide a service beyond what it could with only primary data

References[edit]

  1. ^ "ISWC '19 Conference website". 2019-01-01. Retrieved 2020-01-20.
  2. ^ Euzenat, Jérôme. For Knowledge. 18th International Semantic Web Conference. doi:10.1007/978-3-030-30796-7. ISBN 978-3-030-30796-7.
  3. ^ Henry, Nicholas L. (May–June 1974). "Knowledge Management: A New Concern for Public Administration". Public Administration Review. 34 (3): 189–196. doi:10.2307/974902. JSTOR 974902.
  4. ^ Rowley, Jennifer; Richard Hartley (2006). Organizing Knowledge: An Introduction to Managing Access to Information. Ashgate Publishing, Ltd. pp. 5–6. ISBN 978-0-7546-4431-6.
  5. ^ Atkinson, Rob; Car, Nicholas (26 November 2019). "Content Negotiation by Profile". W3C Working Draft. World Wide Web Consortium (W3C).
  6. ^ Frost & Sullivan (2019-04-18). "Knowledge as a Service (KaaS)".
  7. ^ Yelsky, Simon (2013). "Leveraging Knowledge as a Service to Build a Knowledge Culture in Your Organization". Informa PLC.
  8. ^ "Knowledge as a Service (KaaS)". SURROUND Australia Pty Ltd. 2020.
  9. ^ "Infomill Selected as a Field Service Innovator by IDC : Knowledge-as-a-Service". Infomill. 2017-10-02.
  10. ^ Brenner, Michael (2019-03-20). "Knowledge As A Service: Embracing Disruptors To Drive Revenues". Digital List Magazine.
  11. ^ Anthony J. Rhem (2018-05-15). "Delivering Knowledge As A Service – KaaS". knowledgemanagementdepot.com blog. Digital List Magazine.