Desision Support Systems Essay Example

overall state of the debate. Participants can also modify certain aspects of the system to suit their needs.

Nevertheless, due to the qualitative and publicly controversial nature of public policy argumentation, determining what constitutes an "appropriate design" raises significant issues. The paper aims to provide clarification on the criteria for appropriately designed argumentation systems.In this text, we will discuss two computer-mediated argumentation and decision-making proposals, namely the Zeno system by Gordon and Karacapilidis et al. at the German National Research Centre for Information Technology, and the Risk Agora by McBurney and Parsons. We will also outline the evaluative dimensions suitable for deliberative contexts in which these systems are used. Additionally, we will demonstrate how these criteria can be applied to assess both argumentation systems.

Evaluating argumentation-support systems is crucial in understanding responsible public legitimation's discursive foundations while recognizing formal design limitations in achieving legitimation. With this in mind, we will now focus on two argumentation systems. First is the Zeno system developed by GMD to facilitate decision-making in urban planning as part of a larger European Community-funded project aiming to establish innovative information systems infrastructure for collaborative environmental planning within the public sector.

Zeno's application domain involves multiple stakeholders with diverse backgrounds, interests, preferences, and viewpoints often geographically dispersed. Consequently, the system seamlessly integrates spatially indexed information with non-spatial data and offers intuitive interfaces accessible through an internet platform.Although there were technical challenges in software design, we will not discuss them further. Zeno is a mediation system that focuses on argumentation and decision support features. It is described as a computer-based discussion forum with an emphasis on argumentation. In addition to standard functions like viewing, browsing, and responding

to messages, a mediation system uses a formal model of argumentation. This model helps retrieve and manage dependencies between arguments and provides heuristic information for human mediators in decision-making processes.

Zeno utilizes an adapted formal version of the informal Issue-Based Information System (IBIS) model by Rittel and Webber. This adaptation has been specifically modified for the urban-planning domain. IBIS identifies essential elements of discourse such as Issues, Positions, and Arguments. For example, an Issue could be "Which site should the airport be located?" Positions can then consist of statements indicating different alternative sites or groups of possible sites.

The IBIS model allows discussions to be displayed as hierarchical graphs using hypertext links. This enables users to easily navigate between different threads of the discussion and access supporting documents or contextual information. The goal of the Zeno project is to provide snapshots of debates and easy access to the current state of the planning process.The IBIS model was modified to include preferences, which compare two non-preference positions. Positions in the IBIS model are logical propositions with meaning defined by their role in a specific discussion thread. One approach to siting an airport is considering options such as a "public site" or "easy access". In this case, easy access is considered more important than a public site. This preference becomes a qualitative constraint that may or may not have further arguments or be consistent with other constraints. These preferences and constraints can be discussed and debated, raising arguments and additional issues.

Zeno is a tool that offers users an overview of the argumentative status of positions, preferences, and constraints. It allows users to assess how well each position meets articulated

constraints, labeling them as acceptable or not at any point in the discussion. These position labels can then be combined to assign labels to Issues, indicating the level of argument support based on defined proof standards. The Zeno developers state that there is no universal set of proof standards applicable to all domains. Therefore, they adopted five labels from jurisprudence: (1) Scintilla of Evidence; (2) Preponderance of Evidence; (3) No Better Alternative; (4) Best Choice;The text discusses various systems that aim to support discussions and deliberations. One such system is Zeno, which uses labels with formal definitions based on positions and arguments. These definitions are not legal definitions, but they help show the current status of a discussion and its changes. The designers of Zeno have identified generic speech acts that contribute to a discussion but have not provided a definitive list or rules for their use.

On the other hand, the Risk Agora system, proposed by McBurney and Parsons, fully specifies speech acts and their rules. This system focuses on supporting deliberations on health and environmental risks of new chemicals and substances, as well as their regulation. Typically, these issues involve initial debates within the scientific community about the correlation between potential chemical causes and observed health effects. If a significant relationship is established, further debate occurs among the wider community regarding regulatory options.

Initially, the development of the Risk Agora system has primarily focused on scientific debates (McBurney and Parsons 2000, 2001c).In relation to this matter, the designers have embraced explicit philosophies of science and rational discourse. The philosophy of science they employ is based on Pera's three-person game model of science (Pera 1994). This

model suggests that progress in science occurs through experimentation by scientists, leading to reactions from Nature and ultimately being mediated by the scientific community. The authors assume that members of the scientific community are rational and willing participants in this process. Additionally, they adopt a philosophy of rational discourse for debates within scientific discourses, as assertions in these discourses are open to contestation and defense. To support this viewpoint, they draw upon Habermas's theory of discourse ethics (Habermas 1983/1991), originally formulated by Alexy (1978/1990), as well as Hitchcock's principles of rational mutual inquiry (Hitchcock 1991). Within this framework, the authors explain a dialogue-game inspired by Hamblin (1971) and MacKenzie (1979), outlining pre-conditions necessary for each locution and resulting changes. This dialogue-game allows for assertion, contestation, and defense of propositions, assumptions, consequences, and modes of inference. It essentially functions as a persuasion dialogue (Walton and Krabbe 1995). Some locutions require speakers to defend statements when contested, with the formalism allowing for such defenses to be presented in argument form.Modality can be used to express the confidence of a speaker in their statements, assumptions, consequences, and modes of inference. Commitment stores are utilized in Hamblin games to keep track of assertions made during debates. McBurney and Parsons have developed a system for discussing regulatory options based on Habermas's theory of communicative action. This system defines appropriate types of locutions (speech acts) for such discussions. The formalization of these discussions as dialogue games requires models of deliberation dialogues, which is discussed in another paper at this OSSA meeting by Hitchcock, McBurney, and Parsons. The ongoing work involves specifying the Agora to support regulatory debates. Unlike the Zeno system,

the Risk Agora does not aim to facilitate real-time debates and lacks intuitive graphical interfaces. Instead, it is designed to formally model and represent debates in the risk domain for several purposes: 1) understanding logical implications of scientific knowledge related to a specific issue and arguments on consequences and value-assignments; 2) considering various supporting or opposing arguments for a claim, including regulatory options; 3) analyzing relationships between arguments, certainty levels, strengths, and weaknesses.The objective of the Risk Agora, according to McBurney and Parsons (2001b), is to assist government regulatory agencies in risk assessment and regulatory determination. The Risk Agora is different from Zeno in its goals but shares the aim of capturing snapshots of a debate at any given point. This is especially important in the risk domain, where regulatory decisions must be made before a final determination of scientific knowledge on a specific issue can be reached. Like Zeno, the Agora uses labels to categorize each claim based on presented arguments for and against it up until that moment. Claims in the Agora that have not yet been countered by rebutting arguments or undermined by arguments against assumptions or intermediate premises are considered probable claims. By evaluating the dialectical status of a claim at any given time, the Agora provides an overview of the ongoing debate.The designers of the Agora assess the probability that a snapshot, taken at a certain point after the start of a debate, accurately represents the long-term state of the debate if it reaches stability. This demonstrates how the desirable properties of the Agora can infer longer-term states from finite snapshots (McBurney and Parsons 2001c).

The evaluation and analysis of argumentation

systems can be approached from different angles. For software systems designed for specific, decomposable, and measurable tasks, like generating electricity network bills, assessing system competence and quality is straightforward due to standardized methods that influence good software design (Kirwan and Ainsworth 1993). However, Parker (2000) argues that these methods are not suitable for most decision support systems since decision-making processes cannot be easily reduced to task analysis. She also highlights the lack of developed techniques for evaluating decision support systems. The limited attention given by the Artificial Intelligence community to decision support system quality could possibly be attributed to decisions' resistance to reductionist analysis.

Groothuis and Svensson's (2000) research in the Netherlands focuses on computer-supported welfare assistance decisions. They claim that their investigation is groundbreaking as it examines the extensive utilization of expert systems in handling complex administrative tasks.Designers of knowledge systems, also referred to as expert systems that encode expertise, typically compare the system's performance against a group of human experts using test cases. However, these comparisons pose challenges due to the variability and cultural influences on the definition of "expertise" within different contexts. Moreover, normative decision theory may not accurately reflect actual decision-making processes in certain domains. While this critique could be perceived as a criticism of human decision-makers, it can be difficult to determine whether normative methods are superior in some cases. Additionally, normative decision theories often overlook important non-quantifiable factors crucial for effective decision-making. These complexities underline the inherent difficulties in assessing the quality of advice-giving systems. Is it necessary to conclude that advice is low-quality or unhelpful if it is not followed? How can we evaluate advice when circumstances change between

its delivery and execution? Further complications arise when users disregard limitations or caveats accompanying given advice. How do we assess advice quality in such instances? Lastly, how can we evaluate advice for extreme situations or rare events? These assessments are based on a shared assumption – that there exists a correct or "true" decision attainable by following context-independent inference rules specific to particular domains.However, if this straightforward model encounters problems with knowledge systems, we can expect even greater challenges when evaluating argumentation systems for public deliberation in domains involving multiple parties with complex value-laden issues. For example, decisions regarding urban planning policies may not have inherent truth or correctness that can be accessed independently; any judgment will always be partial and influenced by personal interests. The validity of the decision is determined by the process used to reach it (Forester 1999; Bohman 1996).

Although Zeno and the Risk Agora serve different purposes, they both aim to support deliberative decision-making in a public policy domain with multiple participants. Evaluating these argumentation systems requires an understanding of the relevant type of decision-making, drawing on both argumentation theory and knowledge of public policy formation. Our analysis is based on Forester's deliberative approach (1999, chap.6) to mediated public policy formation and dispute resolution. According to this approach, deliberation is viewed as a transformative and rational process that generates legitimate decisions which are reasonable and publicly acceptable. Four major dimensions emerge from a deliberative approach:This section provides an explanation of these dimensions, which will later be applied to Zeno and the Agora. A deliberative model of policy formation is different from the conventional pluralist model that sees public policy issues as

matters for negotiation and bargaining (Bohman and Rehg 1997). While negotiations are used for resolving decisions involving scarce resources or conflicting interests, deliberation is necessary for making decisions in specific situations. Forester suggests that mediators should approach public planning and policy formation as deliberations, which has significant implications for the process and outcomes. Unlike negotiations where participants aim to compromise their preferences, deliberative political processes require participants to adopt a civic standpoint focused on reaching agreement on a common good or general interest. This is referred to by Forester as "the self-transformative condition." Deliberations provide an opportunity for participants to learn from each other and the interaction itself.Michelman's definition of deliberation involves being open to persuasion by considering the claims of others and one's own reasons. The deliberative medium is a platform for exchanging views, where participants discuss their vital interests and make collective decisions through voting. It is essential for participants to be willing to share information, even if it goes against their self-interests, assuming rationality. If knowledge and preferences are withheld, the decision-making process becomes more like a negotiation than a deliberation.

This approach to deliberation encompasses three dimensions of analysis. Firstly, participants should be open to changing their views based on input from others if it leads to better outcomes - this is known as reasonable self-transformation. Secondly, the properties of topic-specific considerations play an important role in determining conditions for high-quality outcomes. Finally, attention must be given to the participants themselves and their roles within the deliberation process, as well as potential power dynamics that may influence decision-making.

To ensure a satisfactory resolution in disputes or policy matters, it is crucial to consider

specific factors related to the issue at hand. Ideally, all relevant considerations should be taken into account, with particular emphasis on those that are most significant or influential.In an environmental dispute, it is crucial to consider scientific facts about the ecosystem, predictions on the impact of different actions, and economic assessments for affected parties. Disagreements can also arise from conflicting values and fundamental beliefs. The quality of the outcome depends on addressing these relevant considerations during deliberation and processing objections. Relevance is a key concept in this context, although its precise definition is still being researched (Johnson 2000, pp.). There are two conceptions of relevance: one where not considering a consideration can lead to an unsuccessful policy and another where certain considerations like interests and values must be relevant for participants to be overall relevant. This dual understanding leads to a more precise characterization of the substantive quality of a decision based on rational legitimacy and subsequent success rather than reducing it to "truth." Success standards arise from the external aspect of relevance as policy decisions rely on factual assumptions that may be incorrect.An environmental dispute resolution may rely on false scientific assumptions or erroneous economic forecasts, which is why the renewed interest in truth among argumentation theorists is not entirely misplaced. In policy-making contexts, factual claims are intertwined with other types of reasoning. Instead of using terms like "true" or "false," it may be more fitting to use terms like "justified" or "arbitrary". Therefore, participants in policy-making need to consider all relevant factors and construct their viewpoints in a way that is both logically consistent and dialectically responsible or responsive. In summary, being dialectically

responsible means addressing all pertinent questions and objections to arrive at the most plausible outcome compared to other alternatives. Dialectically responsible positions can be connected to a broader understanding of truth. For example, Rescher's conception of plausible reasoning or Hintikka's game-theoretic semantics can be applied (Rescher 1976;1977;Hintikka, 1968). According to these approaches, claims that withstand dialectical reasoning or have the support of a winning strategy in the argumentation system are presumed true but potentially defeasible depending on the burden of proof.This idea of truth is relevant to various decisions in the public policy field, including those pertaining to Zeno and the Risk Agora. It is crucial to address the concerns of all impacted parties and ensure that policy decisions are seen as legitimate. If legitimacy is lacking, compliance with decisions may be adversely affected, underscoring the importance of a deliberative outcome's dialectical quality. The substantive nature of deliberation also prompts consideration of participant inclusiveness and how effectively the process allows different affected parties to express their concerns and influence the outcome. To assess inclusive participation, it is necessary to characterize the involved parties and their roles, including the number of individuals involved in decision-making. The choice of argumentation model used may be influenced by the number of participants, as evidenced by Forester's case studies (1999) on public policy discussions involving multiple parties with diverse perspectives and interests. It is crucial to involve all parties concerned and give them an opportunity to participate in deliberations to ensure legitimacy and subsequent success. However, when evaluating argumentation systems, it is insufficient to only consider explicit exclusions that are clearly defined by roles and entitlements.The importance lies in examining

the hidden forms of exclusion and coercion within the design itself. The system's role in deliberation is crucial as it can either facilitate or impede the expression and reasonable processing of information and viewpoints. This is addressed in the fourth evaluative dimension (3.4), which emphasizes that a legitimate deliberative process should not only allow all affected parties to participate but also provide effective opportunities for them to voice their opinions and enhance mutual learning. Only by doing so can the outcome truly represent an acceptable resolution of the dispute or question at hand. Therefore, this evaluation dimension requires analyzing how well a deliberative design promotes a collective, rational learning process that increases the likelihood of reasonable and generally accepted outcomes, ultimately seen as legitimate. It is necessary to carefully scrutinize deliberation for any subtle forms of exclusion and potential obstacles that hinder participants' capacity to learn from each other while fostering a respectful and inclusive environment. Such analysis surpasses evaluating discussion content and structure alone.Our main objective is to encourage participants to freely express their viewpoints and actively participate in meaningful conversations that promote mutual understanding and the exchange of knowledge.