1 Tor Håkon Sivertsen Bioforsk Plant Health and Plant Protection, Hogskoleveien 7, N-1432 Aas (Norway); e-mail: [email protected] Janis Gailis.

1

Tor Håkon SivertsenBioforsk Plant Health and Plant Protection, Hogskoleveien 7, N-1432 Aas (Norway); e-mail:[email protected]

Janis GailisAgder University College, Service Box 422, N-4604 Kristiansand (Norway); e-mail:[email protected]

A web-based system for administration, verification

and correction of meteorological and biological data in a perspective of actor

network theory

2

Agenda

History

Vocabulary

Design of the system (formal / informal specifications)

Challenges of «web-based» development

Interests of different “stakeholders“

Holistic view: actor-network approach

Perspective: Ubiquity and actor-network breakdowns

3

History

1996: NORPRE project (customization of an existing system for collection of meteorological data from weather stations in Norway)

1996 – 2000: development of the system for administration, validation and distribution of meteorological data (traditional object-oriented systems development methodology with an involvment of different stakeholders at the requirements phase)

2000 – in use at Bioforsk (with modifications)

4

Vocabulary ()

Web-based: a system with human-computer interaction (HCI) implemented by the use of technology of world wide web (Tim Berners-Lee, 1990)

Administration: operations performed by an actor on model instances (in this case instances of meteorological and biological data, and other model elements)

5

Vocabulary

Verification: evaluation (automatic or manual) of model instances according to a set of pre-defined (administrated) rules (facts, predicates, assertions etc.)

Correction: operation (automatic or manual) performed on a specific instance of the model usually based on extrapolation or functional dependecies on other instances in the model

6

Vocabulary

Meteorological data: • based on a classification system (model?) of meteorological phenomena (air, wind, storm, rain etc.)

• attachment of measurable values to phenomena

• metadata (NDFD XML, GIS, WeatherML, SensorML)

7

Vocabulary

Biological data:• also based on a model / theory and describes some biological phenomena (measured or modelled)

• metadata

8

Vocabulary

Actor-network theory: Early 1980's: Centre de Sociologie de L'Innovation

(CSI) École nationale supérieure des mines de Paris

Michel Callon, Bruno Latour, John LawCalled also STS (Science and Technology Studies) scholars,

concerned with qualitative studies of large technical systems; reflects many of the preoccupations of French Post-structuralism but also English-language academic traditions

'Material-semiotic' method. All the elements in a network, human and non-human, can and should be described in the same terms (generalized symmetry).

2006: popular: used to analyze heterogenous relations

9

Vocabulary

Actor-network theory:

Society consists of networks of heterogeneous actors, both human and non-human.

“Actor-Network Theory is useful in the exploration of why technologies, scientific theories, and/or social endeavors succeed or fail as the direct result of changes in their network integrity. In such an analysis, the technology or theory is positioned as the token.“ /wikipedia.org/

10

Vocabulary

Actor-network theory:

“Agents, texts, devices, architectures are all generated in, form part of, and are essential to, the networks of the social.” (Law, 1992)

Sensors (devices), models (architectures), meteorologists and biologists (agents), documentation of the system (texts) are all generated in, form part of, ...

11

Systems development – investigation of the noosphere:

Statement: systems development is a socio-techological process often based on an «ill-defined» problem.

Vocabulary

12

Actor-network theory: punctualization

?

«Workshop on environmental fluid mechanics as elements in agro meteorological modelling»

Punctualization is always precarious,it faces resistance, and may degenerate into a failing network. On the other hand, punctualized resources offera way of drawing quickly on the networks of the social without having to deal with endlesscomplexity. / John Law (1992) /

Vocabulary

13

Actor-network theory:

Bioforsk

Weatherstation

(sensors)

Community of Meteorologists

Model formicroclima

Community of Fluid

Mechanicsresearchers

Vocabulary

14

Actor-network theory:

Core of the actor-network approach:a concern with how actors and organizations

mobilize, juxtapose and hold together the bits and pieces out of which they are composed

Vocabulary

15

16

Design of the system

Requirements• use of data from various sources in different models

• effective retrievement and utilization of data by different actors

• flexible and extendable administration system

17

Design of the system

Parameter

18

Design of the system

Parameter

19

Design of the system

Parameter

QUALITY?

20

Design of the system

Parameters (quality of data)(a) connection of the data to objects of nature («real world» problems? punctualization problems? modelling and scientific method problems?)

(b) properties, conditions and quantitative values of the data, including completeness and representativeness (context dependence ? actor-network dependence?)

21

Design of the system

Parameters (quality of data, cont.)(c) identity of the data that is linked to the social system producing the data (authority, actor-network characteristics)

(d) availability of data (openness? political problems? punctualization problems?)

(e) presentation and use of the data (interpretation) including the context of the presentation (degree of de-punct. ?)

22

Design of the system

Parameters (quality of data, cont.)(f) numerical approximations in digital representation of data (algorithms; hardware; de-punctualization? openness?)

23

Design of the system

Instruments & Sensors

24

Design of the system

Instruments & Sensors

Examples of sensor type (naming convention, typical actor-network breakdown for an actor; de-punctualization):

CP (sensor measuring electric capacity of air) MP-100, Rotronic AG, Germany

CPaa2 (sensor measuring electric capacity of air, second type) HMP45A, Vaisala Oy, Finland

25

Design of the system

Instruments & Sensors

Examples of sensor group type:

ALBSN (measuring 'albedo') and consists of the sensor types TCUPaa4 and TCDO

GRPHDD (measuring 'global radiation') and consists of the sensor type PHDD

26

Design of the system

Instruments & Sensors

Examples of sensor group:

GRTCUPUDN is the SensorGroup measuring global radiation on an instance of a logger (storage) placed at Udnes (could be described as geographical coordinates)

JT10TTHERMUDN is the SensorGroup belonging to the SensorGroupType TTHERM measuring soil temperature in 10 cm depth on an instance of a logger (storage) placed and Udnes (UDN)

27

Design of the system

Coupling instruments and parameters

Example of reification (instruments are «things», parameters are abstractions)

28

Design of the system

Control and correction model

R-test: range test of parameter values that is often connected to the climate at the different sites (existing models)

J-test: jump test (a temporal test comparing values of a parameter at one recorded measurement and the previous recorded measurement; usually depends on existing models)

29

Design of the system

Control and correction model

L-test: consistency test connected to the parameters of one single SensorGroup

LT-test: same as over, but connected to the parameters of two different SensorGroups

CI-correction and CTI-correction are rule-based tests which make up predicates and can be used to make correction of the parameter values provided for the use in the system (more on this at the end of this presentation)

30

Design of the system

Formal language for the validation and correction (to avoid hardcoding)

Lexical issues (permitted characters, reserved words etc.)

Namespaces (models, parameters, sensors, predicates)

Grammar (symbols for objects and relations: model ::= header import* element*)

Precedence and AssociativitySemantics (instances, meaning and relational logic)Types, errors and overloading

31

“Web-based” challenges

• usability (the system should be a part of a [daily] practice of the human agents)

• variety of human agents with different needs in different contexts (heterogenous community)

• accessibility (criteria of openness)• distributed development (no exact knowledge about hardware; unknown infrastructure)

32

Interests of 'stakeholders'

• diverging business interests• power (authority) relations in hierarchic organizational structures

• group psychology • communication competence

33

Actor-networks approach

•'stakeholders' want to keep the network punctualized

•de-punctualization results in conflicts•reification causes ambiguity•PROCESS: problematization, interessement, enrolment, mobilisation of allies (a generalized version of soft systems development methodologies)

•successful interactions of actor-networks as quasi-objects or tokens

34

Perspective: ubiquity ...

•more agents in the system as requirements for ubiquity will dominate more and more

•omnipresence of a sensor network characterizing both local and global environments will become a critical factor for distributed systems systems (GRID)

•integration of actor-networks will be crucial for social relations in the future

•contradiction: depunctualization needed for successful implementation of qualitative systems, but punctualization needed to keep simplicity ...

35

THANK YOU!