Open policy ontology

From Opasnet
Revision as of 15:23, 3 November 2017 by Jouni (talk | contribs) (properties edited)
Jump to navigation Jump to search


The structure of shared understanding describes the information structures that are needed to document shared understanding of a complex decision situation.

Question

What information structures and information tools are needed to document shared understanding in such a way that

  • it can be operationalised and managed and used for automatic inferences by a computer,
  • it can systematically organise information objects used in open assessment, such as variables and statements,
  • can represent each participant's views systematically as a part of the whole even if people disagree,
  • it is intuitive enough to be used by non-experts?

Answer

Shared understanding aims at producing a description of different views, opinions, and facts related to a specific topic such as a decision process. The structure of shared understanding describes the information structures that are needed to document shared understanding of a complex decision situation. The purpose of the structure is to help people identify hidden premises, beliefs, and values and explicate possible discrepancies. This is expected to produce better understanding among participants.

The basic structure of a shared understanding is a network of items and relations between then. Items and relations are collectively called things. Each item is typically of one of the types mentioned below. The first three types are knowledge crystals with the structure question-answer-rationale. They are used in open assessments.

  • Assessment describes a decision situation and typically provides relevant information to decision makers before the decision is made. It is about the knowledge work for decision support.
  • Variable describes a real-world topic that is relevant for the decision situation. It is about the substance of the topic.
  • Method describes how information should be managed or analysed so that it will answer the policy-relevant questions asked. It is about methods.
  • Discussion, or structured argumentation, describes arguments about a particular statement and a synthesis about an acceptable statement. There are two kinds of statements. Note that also the answer of a variable typically is a statement, even if it is often produced by the means of a statistical analysis rather than an argumentation.
    • Fact statement is any text that claims something about the world.
    • Value statement is any text that claims that something ought to be or that it is better then something else.

In the structure of shared understanding, each item may have lengthy texts, graphs, analyses or even models inside them. However, the focus here is on how the items are related to each other. The actual content is here referred to as one key sentence only (description) or even a short, memorable name. However, each item also has a unique identifier (ID) that is used for automatic handling of data. Below is an example of items in a summary table. The whole example discussion can be found from page Voting age.

A sample of item summary from shared understanding about Voting age.
ID Timestamp User Description Type Name
1 Thu Apr 6 13:06:46 2017 Jouni Immaturity is a reason to exclude Value Immaturity
2 Thu Apr 6 13:07:46 2017 Jouni 16-year-olds are mature to do other things as difficult as voting sex, army, speaking publically Value Mature enough
3 Thu Apr 6 13:06:46 2017 Jouni Most 16-year-olds are not mature enough emotionally Fact Not yet mature
6 Thu Apr 6 13:09:46 2017 Jouni Actions that reduce political apathy should be implemented Value Reduce apathy
7 Thu Apr 6 13:10:46 2017 Jouni Waiting before voting increases apathy Fact, variable Waiting passivates
8 Thu Apr 6 13:11:46 2017 Jouni Young people are politically apathic anyway. Fact Young apathetic
27 Thu Apr 6 13:06:46 2017 Jouni 16-year-olds are mature enough to vote Value 16 are mature
4 Thu Apr 6 13:06:46 2017 Jouni If you are taxed, you should get to vote Value Voting if taxed
5 Thu Apr 6 13:06:46 2017 Jouni Everyone who uses money pays VAT tax. Fact Everyone is taxed
37 Thu Apr 6 13:06:46 2017 Jouni Voting right can be given based on specified reasons Value Given with reasons
A sample of relations about Voting age. For descriptions of subjects and objects, see the table of items above.
ID Timestamp User Description Subject Predicate Object Name
6 Thu Apr 6 13:06:46 2017 Jouni Opposes = if – then not Not yet mature opposes Mature enough
7 Thu Apr 6 13:06:46 2017 Jouni Supports = if – then Mature enough supports 16 are mature
8 Thu Apr 6 13:06:46 2017 Jouni Immaturity AND 16 are mature supports Voting for 16
11 Thu Apr 6 13:06:46 2017 Jouni Given with reasons AND Voting if taxed supports Voting for 16
14 Thu Apr 6 13:06:46 2017 Jouni Voting if taxed makes relevant Everyone is taxed

Rationale

Ontological structure

Resource has subclass

  • Item
  • Property

Item has subclass

  • Assessment. Assessment has parts
    • Question
    • Answer
    • Rationale
  • Variable
    • Question
    • Answer
    • Rationale
  • Method
    • Question
    • Answer
    • Rationale
  • Discussion. Discussion has parts
    • Statement. Statement has type
      • range: {Value statement, Fact statement}
    • Resolution
    • Argumentation
  • Action. Action has parts
    • Who
    • When
    • What

Properties

Relations are sentences that connect one item (subject) with a verb (predicate or property) to another item (object), possibly with qualifiers about scoping, references etc. In the structure of shared understanding, one key idea is that the relations are straightforward and simple enough so that a computer can make inferences about the items and their relations. Therefore, the number of relations is kept small. Some relations are the same as used in Wikidata. This approach is compatible with resource description framework (RDF), the same system that is used in Wikidata database. These are the key relations:

  • From set theory:
    • instance of (<> inverse of has instance): subject belongs to a set defined by the object and inherits the properties of the set.
    • subclass of (<> has subclass): subject is a subset of object.
  • From logics:
    • if - then: If subject is true, then object is true. Also the negation is possible: if - then not.
    • opposite of: subject is opposite of object.
    • and, or, equal, exists, for all: logical operators.
    • has context: subject given that object is true.
  • Causal:
  • Other:
    • makes relevant (<> is relevant given): if the subject is relevant in the given context, then also the object is. This typically goes from a variable to value statement or from a value statement to a fact statement.
    • has reference (<> is reference for): object is a reference that backs up statements presented in the subject.
    • has tag (<> is tag for): object is a keyword, type, or class for subject. Used in classifications.
    • associates to (<> associates to, i.e. is its own inverse): subject is associated to object in some way. This is a weak relation and does not affect the outcomes of inferences, but it may be useful to remind users that an association exists and it should be clarified more precisely.
    • has truthlikeness: A subjective probability that subject is true. Object is a numeric value between 0 and 1. Typically this has a qualifier "according to X" where X is the person or archetype who has assigned the probability.
    • has preference: subject is better than object in a moral sense.
    • attacks, defends, comments: these argumentative relations are typically used within discussions to reach a statement. Therefore they rarely show up in the structure of shared understanding.
    • There may also be mathematical and functional relations, but they are typically used in models within variables. Therefore they rarely show up here.
    • exists within (the subject exists within the boundaries set by the object; existence does NOT mean that subject is true, only that there is a reason to consider its contents.
    • There are important relations between decisions. Actions may be exclusive (if you do one, you cannot do another). Some sets of actions may be mutually exhaustive (all possible actions in a given situation are listed). It should be possible to describe these issues with relations as well, but such properties are not well developed yet.

Free-format information typically has following properties:

  • has topic (<> is discussed in)
  • relates to stakeholders (<> is involved in?). There may be a need for more detailed stakeholder properties:
    • participates in
    • affects (a particular decision process)
    • has interest in
    • finds important
    • claims
  • associates with (a weak relation)
  • answers question (<>is question for)
  • has category (<> is category for)
  • has owner (<> is owned by)
  • has summary (<> is a summary for)
  • has reference (<> is a reference for)

Topic, question, category, and summary are more structured and more precisely defined objects than a free-format object, and therefore it is more useful to use those in further information strucures rather than the original document. ----#: . These properties are fairly similar and possibly we don't need all of them. --Jouni (talk) 06:25, 30 October 2017 (UTC) (type: truth; paradigms: science: comment)

Structured objects typically have following properties:

  • is instance of {fact, value} (facts and values are subclasses of claims)
  • one of logical properties (see above)
  • one of causal properties (see above)
  • is relevant for ----#: . Is this the same as 'makes relevant'? --Jouni (talk) 06:25, 30 October 2017 (UTC) (type: truth; paradigms: science: comment)
  • is irrelevant for
  • has truthlikeness
  • has popularity
  • has value (normative)

Open assessments typically have following properties:

  • is instance of {assessment, variable, method}
  • has causality {function, conditional probability}

Relevant information areas

The five information areas used in shared understanding (open policy practice ontology). If a piece of information does not fit into any of these areas, it is irrelevant for the decision at hand. For description of the content of each area, see text.

This ontology is specifically about decision making, and therefore actions (and decisions to act) are handled explicitly. However, any natural, social, ethical or other phenomena may relate to a decision and therefore the vocabulary has to be very generic. When we descbribe actions under planning, those descriptions are called "Decision processes", while all other descriptions are called "Substance".

The properties described below are used in the following kind of sentences: "Information X describes Y of decision Z" where X is an identifier of a particular piece of information that describes a phenomenon, Y is the information area that is described, and Z is an identifier of a particular decision.

  1. Substance (information about a substantive topic or phenomenon itself): What issues relate to a decision Z? What causal connections exist between issues? What scientific knowledge exist about the issues? What actions can be chosen? What are the impacts of these actions? What are the objectives and how can they be reached? What values and preferences exist?
  2. Decision process (information about how a decision Z will be made): What will be decided? When will it be decided? Who has the authority to decide? Who are involved? How is the decision prepared? What political realities and restrictions exist?
  3. Tasks (information about organising the information work to support decision making): What tasks are needed to collect and organise the information? When do these tasks need to be done? Who is responsible of what? How is information work organised? Tasks are also important afterwards to distribute merit and evaluate the process: Who did what? How did information evolve? Where did data come from?
  4. Methods (information about methods used in the information work): How to perform information work? What methods are available for a task? How to participate in the work? How to use statistical and other methods and tools?
  5. Participants (information about participants and decision makers in the information work): Who participates? Who should participate? Who has necessary skills for contributing? How to motivate participation? How to measure merit of contributions?
  6. Irrelevant issues: Information that do not fall into any of the previous categories is thus irrelevant for a particular decision Z. If there is no identified relation between an information object and a decision, it implies that the information object is irrelevant. However, because not all relevant relations have been considered and documented, it is often useful to explicate the irrelevance, especially if people may (falsely) think that it is relevant.

How to manage shared understanding

Shared understanding is a structured description of a decision situation. A key idea is that it is much faster to produce than a quantitative assessment, is more usefully organised than a free-format document (not to mention unmoderated discussions), and can distill information from both into a coherent information structure.

Questions for further developing shared understanding:

  • What are the main questions within each property?
  • What are necessary structures and relations?
  • What software tools can be used?

Example of using the structure

Columns that may have several values per risk are marked with *

Riskilomake (a variable with the question: What is a risk that is relevant for the success of THL's mission? There are several variables with an identical question, but each variable describes exactly one risk as an answer.)

Each column is described within the variable answer unless otherwise noted a property that is used to link the column contents to the variable.

id# Tarkastelukohde (yksikkö) * (has tag) Riskialue (aihepiiri karkea) * (has tag) Sisäalue / sisältösivu (aihepiiri tarkka) * (has tag) Riski Tarkennus Todennäköisyys Vakavuus Riskiluku (todennäköisyys*vakavuus) Hallintatoimet (kpl) Muistiinpanot Omistaja Tila Hallintatoimien valmius-% Omat liitteet * Luotu (automatic from version control) Päivitetty (automatic from version control)

Hallintatoimilomake (many-to-many relationship with risks) id# Hallintatoimi Määräpäivä Vastuuhenkilö Tila Omat liitteet * Luotu (automatic from version control) Päivitetty (automatic from version control)

Deliberative democracy

James Fishkin, a key proponent of deliberative democracy, describes two approaches to public opinion, raw vs. refined: what people actually think vs. what their opinion would be after it has been tested by the consideration of competing arguments and information coscientiously offered by others who hold contrasting views. Political process can be seen as whether a filter or a mirror. The filter creates counterfactual but deliberative representations of public opinion. The mirror offers a picture of public opinion just as it is, even if it is debilitated or inattentive. The conflicting images suggest a hard choice between the reflective opinion of the filter and the reflected opinion of the mirror.[1]

It is only through the deliberations of a small face-to-face representative body that one can arrive at the "cool and deliberate sense of the community" (James Madison, Federalist No 63). ... A key desideratum in the Founders' project of constitutional design was the creation of conditions where the formulation and expression of deliberative public opinion would be possible.[1] A smallish group of randomly selected people are likely to act as a filter, while e.g. a referendum would act as a mirror. During the early days of the United States, James Madison actively designed governance structures that would enable the formation of refined public opinion in the national US policy. The electorate was such a construct, designed to enable informed argumentation about president candidates before the final vote. However, this role has completely disappeared, as nowadays the outcome of the electoral vote is known as soon as the composition of the electorate is known.

Shared understanding follows these lines of reasoning and aims to produce a deliberative outcome of informed argumentation. However, the major difference is that the deliberative process does not aim to produce a decision by the participants, but a comprehensive description of shared understanding with all relevant points and disagreements. This written description enables other people to learn and form their own opinions of the matter, and thus help in other similar decision situations. Although producing such a description may be time-consuming and labourious, re-usability of the information makes it worth the effort.

Cognitive democracy

  • Henry Farrell (George Washington University), Cosma Shalizi (Carnegie Mellon University and The Santa Fe Institute). 2012?. An Outline of Cognitive Democracy [1],

Farrell and Shalizi analyze three main approaches to socially achieve results: hierarchies in different forms (with problems that those who are in power are not receiving information from the others); markets (with problems that they converge to individual benefit, which is sometimes in conflict with social benfit), and democracy (with problem how to actually implement the main principle of equal power among individuals). They suggests approaches to improve democracy.

Pol.is

Pol.is is a website for organised democratic discussion. It helps large organizations and communities understand themselves by visualizing what people think.

  • An example discussion about sote indicators [2] (in progress)
  • A case study of temperature check [3]
  • A case study from Taiwan [4]: vTaiwan: Public Participation Methods on the Cyberpunk Frontier of Democracy. In the midst of the signal failure known as the US electoral season, here’s something to be inspired about: a true story about rational deliberation on a national scale.

Professionalism

Jonathan Rauch and Benjamin Wittes. (May 2017) More professionalism, less populism: How voting makes us stupid, and what to do about it. Center for Effective Public Management at Brookings. [5]

Artificial intelligence

Rauhankone

Artificial intelligence may solve some of the structural problems related to development of shared understanding. How this would actually happen is largely unclear. However, professor Timo Honkela is working toward this aim. For more details, see op_fi:Rauhankone.

Inforglobe

A similar but simpler approach is by Mikaeli Langinvainio and Juha Törmänen, who used to work for Crisis Management Initiative. They use statistics to understand views and opinions of different stakeholder groups. (HS 25.6.2017 Voiko rauhanneuvotteluja edistää matematiikalla?) Their company inforglobe produces consulting services based on these ideas. [6]

Their web tool contains these information structures and functionalities (for more details, see Inforglobe link above):

Likelihood of affecting the project vs impact for the project vs knowledge level on the risk or threats vs opportunities

Attributes

  • Categories (e.g. project planning, logistics and safety, or joker risks)
  • Participants (e.g. project team, planning organisation, partner organisation, or customer representative)

Issues e.g.

  • Contractor network (joker)
  • Cost stucture (planning)
  • Logistics
  • Machinery placement (logistics and safety)
  • Staff competence (planning)

Values e.g.:

  • Large or small 1-5
  • Likely 1-5
  • Knowledge good 1-5

Additional properties

  • Each value can be enhanced with a suggestion how to mitigate the impact or decrease the likelihood.
  • Individual answers can be shown with participant attributes and suggestions.
  • Each issue has a more detailed description.
  • Based on individual answers, you can make shared conclusions about each issue and how to manage the risks.
  • Assessment can be done several times.
  • Participants can be categorised based on position or sector (and maybe other attributes as well)

System dynamic maps

  • Issues as nodes
  • Complex system maps: Causal edges between them with strength
  • Significance of edges (links) is measured in some way and used to select nodes and edges for display
  • Co-operation: Links can also describe how well items (in this case organisations) communicate with each other.

See also

References

  1. ↑ 1.0 1.1 James Fishkin. (2011) When the people speak. Democratic deliberation and public consultancy. Publisher: Oxford University Press. ISBN 978-0199604432
Retrieved from "https://en.opasnet.org/index.php?title=Open_policy_ontology&oldid=41459"