Scavenger: More information

The main aim of the Scavenger approach is to support the investigation, selection, structuring, sorting, composing and organising of fragments of data from original source materials. Such original sources include electronic manuals, paper based manuals, photos, handwritten notes, paper artifacts, recorded audio and so on. Scavenger is particularly generic and may be utilised in a variety of domains providing that pre-existing original sources in either electronic or physical format are available. It is however important to note that all physical sources must be digitised in some way so that they may be manipulated by Scavenger.

Scavenger provides a variety of generic data acquisition mechanisms for various types of media to support the recovery of fragments of data from existing artifacts. These mechanisms are constructed in such as way that they may support modelling for a variety of different applications. The output of Scavenger is an open, structured database that may be subsequently accessed for further interrogation and processing. A key feature of Scavenger is that it is model independent and can be used in the construction of a wide variety of entity-relations based models. So that this database may be of most utility, it's format and structure is specified by the "customer" application that is to make use of the data after scavenging has taken place. Subsequent usages of the data models generated by Scavenger could involve browsing by humans, advanced querying, data re-factoring, automated analysis and so on.

We believe that the suitable structuring and presentation of ethnographic information is essential to support the process of qualitative analysis. In particular, the facilities provided by Scavenger aid both investigation and comprehension by ethnographers, system designers, engineers, managers and other interested parties. In addition to this, much can be gained from the pre-processing of raw ethnographic data in order to distil, refine and restructure it into an appropriate format to support direct access by other existing software tools and methods.

Scavenging is the process of selecting fragments of media in order to create and populate entities and their associated elements. In order to achieve this, a number of different media selectors are used:

Textual media selector - Allows the user to select fragments of electronic text (must be store as plain ascii or html).

Graphical media selector - Allows the user to select fragments of images

Audio media selector - Allows the user to select segments of audio

The creation of new entities can occur once one or more fragments of media have been selected. The user then simply selects the desired entity type and a new entity of that type is created, with all the selected fragments of media being entered as elements of that entity's name. All fragments which have been used in this way, then become labelled as previous selections. All entities that are created by this process will contain references back to the fragments of raw data from which they were created.

A user can populate entities by "Scavenging" media fragments for the entity name and description. This is done by selecting new fragments in the media selectors and then requesting the selected text to be imported. It is then always possible to enter extra text manually for the entity name and description. Users can also setting description flags for a particular entity to aid in defining it's exact nature. Relationships between entities can be created by selecting the desired target from list of existing entities. A bi-directional relationship is then created between the entity being edited and the target entity which was selected. Finally, relationship lists can also be reordered, if ordering is important for a particular model.

LINKS

Strider

Situation modelling

Example: Strider

One example of the adoption of Scavenger is for data reclamation within Strider configuration models. The aim of Strider is to support the modelling and analysis of the top-level configuration of socio-technical systems. The focus of Strider is on the composition and structure of a system's configuration, rather than the operation of the system itself. The main aim is the modelling and dependability analysis of the configuration that currently pervades in an existing socio-technical system.

In the Strider approach, configurations consist of components and the processes that related those components together. Components can represent a wide variety of phenomenon found in socio-technical systems, ranging from pieces of software and hardware, through buildings and physical spaces, to the people who work in an organisation. The Strider models are based around two separate perspectives of a system, structural model (demonstrating "part-of" relationships) and a set of process models (demonstrating abstract "used-by" relationships). The Strider entities that Scavenger can be used to construct include the following:

o Configuration Item - A component which forms part of a system configuration
o Class - Defines a particular class of configuration item which may be present in a system configuration
o Process - A work process which relates a number of configuration items together
o Composite - A commonly experience, complex configuration "cluster" (focusing on structure) which is constructed on multiple atomic configuration items
o Pattern - An abstract commonly experience, complex configuration "cluster" (focussing on process) which involves multiple atomic configuration items
o Objective - A high level task or aim the attainment of which is supported by the system configuration
o Weakness - An identified problem or anomaly associated with a process which may prevent the achievement of objectives

The role of Scavenger within the overall Strider approach is to identify, describe and structure key system components that are derived from an input set of ethnographic source materials. Scavenger also permits relationships to be specified between entities (e.g. sub-class, instance-of, part-of, has-input etc). In order to achieve the Scavenging of components and their formatting into an appropriate data structure, an XML template has been specified for each of the previously mentioned entities.

The Strider tool operates directly on the produced Scavenger database, utilising a standard XML parser to obtain access to the data held within. Based on this data, Strider is able to produce graphical representations of both the structural and process models. The Strider tool also allows the user to make alteration and additions to the data originally acquired by Scavenger. Any changes made to the data are written back to the database in the form of XML documents that are consistent with Scavenger format. The Strider tool augments the abilities of Scavenger to provide more specialised graphical editing facilities for the specification of components and processes.

LINKS

Strider configuration modelling and analysis approach

Example: Situation Modelling

One example of the use of Scavenger focuses on its adoption for data reclamation in Situation Modelling. Situation modelling can be thought of as the sophisticated modelling of environments and provides a structured ethnographic description of a particular work situation. The aim is to provide a way of presenting situation information to designers in order that they can gain a better understanding of the system requirements and various domain issues. In this way situation modelling attempts to inform the design process and improve the quality of the finally produced system. Situation modelling allows designers to get first hand experience of the context in which their system will be deployed. It also provides a way for ethnographers to communicate their findings to a wider audience and at the same time provide them with a tool that will aid in the compilation of reports.

A situation model should provide an overall view of the situation from the basic hierarchy of elements, to very detailed information about specific tasks. This is a great deal of information to accommodate in one model and so the notion of perspectives or viewpoints are needed. There are three main viewpoints, the Task viewpoint that looks at large granularity work tasks, the Role viewpoint that looks at operational roles within an organisation, and the Artifact viewpoint that looks at artifacts from the application domain. The entities that form these viewpoints are listed below, some are shared between viewpoints and others are unique to a particular viewpoint:

o Stream - A stream is an ongoing series of activities that has some purpose, with no clear beginning or end.
o Activity - An activity is a set of defined/recognisable units of work with a specific purpose. It is a unit that has a beginning and an end, but can be made up of sub activities. Activities are also made up of events and may be ordered into coherent sets called Event sets. These may be carried out one or more times.
o Event Set - An Event set contains smaller events that must be carried out in order to complete the activity. The events may need to be carried out in a specific order.
o Sequence - An ordering or partial ordering of events or activities.
o Event - These are units of work that can be observed/identified and have an observable beginning and an end. They are atomic.
o Explanation - This is an explanation of the event from a certain point of view. There may be more than one explanation for a single event.
o Vignette - These are short examples, or instances of the events; they can be used to further explain the event, or a viewpoint.
o Placement - The placement defines where the role is within the hierarchy of the situation.
Interaction - Interactions describe the actors and artefacts that this particular role encounters.
o Starting Environment - The starting environment describes where the artefact begins, where it starts off life.
o Setting of use - Where the artefact is required and by whom. This highlights the interactions between the roles, actors and various tasks.

Scavenger can be used to construct entities of all of the above types by performing reclamation of data from ethnographic original sources such as photographs, domain artifacts, audio records, handwritten fieldwork notes, electronic reports etc. In this role Scavenger is basically being used to add additional form and structure to the original ethnographic information. By specifying an XML template for each of entity type, a Scavenger database schema can be created which conforms to the data model mandated by the Situation Modelling approach. The entity classes specified using such templates and the subsequent entity instances created and populated by Scavenger form the core to the database that is produced by data reclamation. Situation modelling can then utilise this database to help inform the design decisions of system developers, managers and so on.

Scavenger handles all creation, management and alteration of the entire structured models upon which Situation Modelling is based. The activities that are carried out upon the Scavenger database after scavenging has taken place are all investigation, explanation and discussion oriented tasks performed by human operatives. Although no automated activities are performed upon the Situation Models produced by Scavenger, a support tool is planned which will include a variety of querying and reporting mechanisms which will operate upon this data.