108
R. Pandey and M. Pande
or processes. A workflow is signified as chaining of various activities together. An
activity A2 may be taking input data from an activity A1. Thus in workflow provenance, the emphasis is on activity rather than the data that is being worked upon.
Whereas in data provenance the emphasis is on the data. Computationally-driven
data-intensive experiments using workflows enable automation, scaling, adaptation,
and provenance support [14]. In workflow provenance, the emphasis is on the procedure/workflow rather than the data. Thus, workflow analysis can enable evaluating
the degree of provenance.
4 Provenance Models
Models of provenance evolved since there was no formal way to represent data about
provenance in the World Wide Web and IoE. Hence considering the above concern,
the model of provenance was first proposed in 2006 when a common consensus
emerged among the members of the W3C consortium [15]. A common model of
provenance was prescribed for representing provenance due to the following reasons.
• need for a common vocabulary of provenance representation,
• uniformity in the agent structure and functionality,
• documentation of process related to provenance,
• how of data derivation, is data item the result of an activity or an agent,
• syntax of data annotation.
Subsequent to the consensus for the uniform model, a challenge was proposed by
the W3C which led to the development of OPM or the open provenance model for
prescribing provenance [16].
4.1 OPM Model Leading to PROV-DM
The OPM model based on the principle of causal relationships emphasizes, the
capturing dependencies evolving out of the processes.
It is further stated that these dependencies can be captured by capturing the
processes through which the artifacts were derived using these.
The OPM model studies causal relationships or dependencies on processes by
which the artifacts were derived and the Agent that initiated these processes. For
example, a book B was produced by the editing process E and by the help of agent
A and subsequent modifications/additions to book B. It also allows us to assert the
derivation dependency between different artifacts without specifying any processes
that were involved in creating that artifact. It is thus inferred that agents can be human,
machines, or organizations that might be involved in the creation of an artifact with
the help of a process.
R. Pandey and M. Pande
or processes. A workflow is signified as chaining of various activities together. An
activity A2 may be taking input data from an activity A1. Thus in workflow provenance, the emphasis is on activity rather than the data that is being worked upon.
Whereas in data provenance the emphasis is on the data. Computationally-driven
data-intensive experiments using workflows enable automation, scaling, adaptation,
and provenance support [14]. In workflow provenance, the emphasis is on the procedure/workflow rather than the data. Thus, workflow analysis can enable evaluating
the degree of provenance.
4 Provenance Models
Models of provenance evolved since there was no formal way to represent data about
provenance in the World Wide Web and IoE. Hence considering the above concern,
the model of provenance was first proposed in 2006 when a common consensus
emerged among the members of the W3C consortium [15]. A common model of
provenance was prescribed for representing provenance due to the following reasons.
• need for a common vocabulary of provenance representation,
• uniformity in the agent structure and functionality,
• documentation of process related to provenance,
• how of data derivation, is data item the result of an activity or an agent,
• syntax of data annotation.
Subsequent to the consensus for the uniform model, a challenge was proposed by
the W3C which led to the development of OPM or the open provenance model for
prescribing provenance [16].
4.1 OPM Model Leading to PROV-DM
The OPM model based on the principle of causal relationships emphasizes, the
capturing dependencies evolving out of the processes.
It is further stated that these dependencies can be captured by capturing the
processes through which the artifacts were derived using these.
The OPM model studies causal relationships or dependencies on processes by
which the artifacts were derived and the Agent that initiated these processes. For
example, a book B was produced by the editing process E and by the help of agent
A and subsequent modifications/additions to book B. It also allows us to assert the
derivation dependency between different artifacts without specifying any processes
that were involved in creating that artifact. It is thus inferred that agents can be human,
machines, or organizations that might be involved in the creation of an artifact with
the help of a process.