This paper investigates the evolution from a medical appointment scheduling framework based on Answer Set Programming (ASP) integrated with Blueprint Personas to a more cognitively rich, agent-based scheduling system employing the L-DINF epistemic logic framework. We illustrate how agent-oriented models incorporating beliefs, intentions, and dynamic reasoning capabilities can efectively enhance or replace the persona-based constraint optimization traditionally used. Key advantages of the L-DINF model, such as improved adaptability, enhanced explainability, and more human-like decision-making, are emphasized. Furthermore, a structured translation methodology from static personas into dynamic epistemic agents is proposed, accompanied by a modular logical architecture supporting real-time, responsive scheduling.
plexity of balancing limited resources, patient urgency, and individualized preferences. Traditional
manual scheduling methods and simple heuristic-based solutions [
and combinatorial optimization [
become evident in dynamic settings. Personas are, by design, static abstractions and do not support real-time reasoning, belief updates, or proactive behavior. As a result, they cannot easily accommodate changing availability, evolving preferences, or unforeseen disruptions in clinical operations.
To address these limitations, we explore the integration of L-DINF, a logic-based framework for
modeling intelligent agents with epistemic capabilities such as beliefs, intentions, preferences, and
contextual reasoning over actions and environmental changes [
ceur-ws.org
agents, articulates the rationale for the integration of L-DINF into ASP-based systems, and demonstrates how such integration addresses the limitations of purely static models. We argue that this hybrid approach is particularly well-suited to the healthcare domain, where scheduling must be responsive to constant change.
The paper is organized as follows: Section 2 reviews the Personas and L-DINF frameworks; Section 3 elaborates on the motivation to integrate epistemic agents into persona-based scheduling; Section 4 analyzes the feasibility of this integration; Section 5 details the translation of personas into L-DINF agents; Section 6 provides a running example; and Section 7 summarizes our findings and outlines future research directions.
archetypes representing prototypical patients. They combine clinical information (e.g., chronic
conditions), social context (e.g., dependency on caregivers), cognitive attributes, and digital literacy levels
[
our system should be understood as abstract ontological templates, conceptual structures that define variables and relationships related to the patient. These templates are instantiated using real patient data, EHR (Electronic Health Records), and system-level parameters, which populate the factual layer of the ASP model.
Each persona defines a structured combination of clinical status, socio-environmental context, and digital capabilities. This layer of abstraction allows the system to reason over complex, human-centered scheduling needs without hard-coding per-patient logic. Specifically, a patient persona may include attributes such as: personal identity and geographical location, physical or mobility conditions (e.g., disability status), clinic preferences or accessibility needs, sensory sensitivities (e.g., noise or light sensitivity), preferences over physician specialization and experience, preferred time windows for appointments and distance or travel time to clinics. These high-level profiles are encoded as ASP facts, forming the basis for reasoning: 1 patient(p1, "Mario", "Rossi", "L'Aquila"). 2 disabled(p1). 3 preference(p1, c3). 4 sensory_preference(p1, "noise"). 5 doctor_preference(p1, "GP", "chronic_diseases", 10). 6 appointment_preference(p1, c3, 1850, 2000). 7 distance(p1, c3, 15).
Clinician personas are modeled in the same way. They include attributes such as medical expertise, experience, and operational limitations, enabling the system to consider stafing constraints and match appropriate providers to patient needs, in fact each visit is identified by: name, cost and classification indicating its chronicity (0 = non-chronic, 1 = chronic): 1 doctor(m1, "Marco", "Bianchi", 52, "L'Aquila", "GP"). 2 doctor_experience(m1, "GP", 25). 3 doctor_experience(m1, "chronic_diseases", 5). 4 5 visit_type(v1, "Cardiology", "Heart Attack", 0, 0, 0). 6 visit_cost(v1, 1000). 7 required_sessions(v1, 2). 8 session_interval(v1, 14, 28).
in a way that is both medically sound and personalized. It also enables scenario-based validation, where synthetic patients are simulated to test how the system handles edge cases or vulnerable populations.
guiding the solver toward optimal, patient-centered outcomes. Preferences are modeled as soft rules, contributing weighted terms to the objective function.
For instance, a patient’s preference for a particular doctor type is captured by a scoring rule: 1 appointment_preference_effect(Patient, Time, Clinic, 1) :2 patient(Patient, _, _, _), 3 clinic(Clinic, _), 4 availability(Clinic, _, _, Time), 5 appointment_preference(Patient, _, Start, End), 6 X = (((Time \ 86400) * 3600) * 100) + (((Time \ 3600) / 60) / 3) * 5, 7 X <= End, X >= Start.
operational, and ethical requirements. These constraints ensure the feasibility of assignments.
1 Sessions { appointment(Patient, Clinic, Doctor, Visit, Time) : 2 availability(Clinic, Doctor, Visit, Time) } Sessions :3 need(Patient, Visit, _), 4 required_sessions(Visit, Sessions).
delivery modes—are implemented to ensure system realism. For full access to the rules, we provide our codebase online1.
The logical framework L-DINF, that we illustrate in this subsection, allows the modeling of group
dynamics of cooperative agents. Consequently, one can model agents that can form groups and support
each other in performing collective mental actions [
possibility for agents to have roleswithin their group of agents. Roles determine which actions each
agent is enabled by its group to perform [
can maintain a version (possibly outdated) of the mental state of other agents and perform inferences about such knowledge.
for the logic’s core, and it is proven to be strongly complete, but that does not guarantee computational
tractability, reasoning in such a rich system is PSPACE-hard; for more detail refer to [
L-INF, is a logic of explicit beliefs and background knowledge. The second component extends the static one with dynamic operators that express the consequences of agents’ mental actions. 2.2.1. Syntax
Let Atm = {, , …} be a countable set of atomic propositions. The set is the set of the physical actions that agents can perform, including “active sensing” actions (e.g., “let’s check whether it ra,ins” “let’s measure the temperatu, reet”c.). Let Agt be a set of agents and Grp the set of groups of agents.
The language of L-DINF, denoted by ℒL-DINF, is defined by the following grammar:
pref _do ( , ) ∣ pref _do (, ) ∣ . , ∶∶= ∣ ¬ ∣ ∧ ∣
B ∣
K ∣ do ( ) ∣ doG ( ) ∣ can_doG ( ) ∣ intend ( ) ∣ exec () ∣ [ ∶ ] ∣
Cl( , ′ ) ∣ fCl ( ) ∶∶=
+ ∣ ⊢(, ) ∣ ∩(, ) ∣ ↓(, ) ∣ ⊣(, ) defined from form [ ∶ ] where ranges over Atm, , ′ ∈ ¬ and ∧ in the standard manner.2 The language of mental actionosf type is denoted by , ∈ Agt, ∈ ℕ , and ∈
ℒACT. The static part L-INF of L-DINF, includes only those formulas not having sub-formulas of the
are interested in modelling the reasoning of agents acting cooperatively. We consider the set of agents as partitioned in groups: each agent ∈ Agt always belongs to a unique group in Grp. We assume that all agents initially belong to an initial group. Any agent , at any time, can perform a (physical) action joinA(, ) , for ∈ Agt, in order to change her group and join ’s group. The special case in which = denotes the action that allows agent to leave her current group and form the new singleton group {} .
The formula intend ( ) indicates the intention of agent to perform the physical action , in the
sense of the BDI agent model [
we will write as subscript in place of {} . So, for instance, we often write exec ( ) instead
The formula doi( ) indicates the actual executioonf action by agent, automatically recorded by the new belief do ( ) (postfix “ ” standing for “past” action). Note that, we do not provide an axiomatization for do (and similarly for doG , that indicates the actual execution of by the group of agents ). In fact, we assume that in any concrete implementation of the logical framework, do and doG are realized by means of a semantic attachmen[t20], that is, a procedure which connects an agent with its external environment in a way that is unknown at the logical level. The axiomatization only concerns the relationship between doing and being enabled to do.
The expressions can_doi( ) and pref _do ( , ) are closely related to doi( ). In particular, can_doi( ) must be seen as an enabling condition, indicating that the agent is enabled to perform the action , while pref _doi( , ) indicates the level of preference/willingness of agent to perform .
The formula pref _doG (, ) indicates that agent exhibits the maximum level of preference on performing action within all group members. Notice that, if a group of agents intends to perform an action , this will entail that the entire group intends to do , that will be enabled to be actually executed only if at least one agent ∈ can do it, i.e., it can derive can_doi( ).
The formula Cl( , ′ ) denoted the equivalence of the two physical actions and ′ . Intuitively, this means that in the specific practical context at hand, the two actions have “something in common”, i.e., for instance, they use similar resources, perform in a similar way, can be used by an agent to obtain equivalent results, etc. Notice that the predicate Cl induces a partition of in a collection of equivalence classes.
to represent beliefs, i.e., facts and formulas acquired via perceptions during an agent’s operation, and a long-term memoryused to model agent’s background knowledge. Such knowledge is assumed to satisfy omniscienceprinciples, such as: closure under conjunction and known implication, closure under logical consequence, and introspection.
Background knowledge of an agent is specified by means of the modal operator K , which is actually the usual S5 modal operator often used to model knowledge. The fact that background knowledge is closed under logical consequence is justified because we conceive it as a kind of stable and reliable knowledge bas.eThe modal operator B , instead, is used to represent the beliefs of agents kept in ’s working memory. The contents of the working memory is determined by the mental actions has executed. We assume the background knowledge to include: facts/formulas known by the agent from the beginning, and facts the agent subsequently decided to store in its long-term memory (via a decision-making mechanism not covered here) after processing them in its working memory. We therefore assume that background knowledge is irrevocable, in the sense of being stable over time.
Whenever an agent wants to perform a physical action ′ , it can exploit the equivalence described by the facts of the form Cl( , ′ ) to execute a most convenient action (in terms of resources requires, preferences, etc.) drawn from the equivalence class of ′ . The formula fCl ( ) indicates that is the more convenient action among those in the set { ′ |Cl( , ′ )}.
The formulas exec () express executability of mental actions by a group (which is a consequence of the fact that any member of the group is able to perform the action). They have to be read as: “ is a mental action that an agent in can perform”.
has been performed by at least one of the agents in , and all agents in have common knowledge about this fact”.
Let us now introduce the dynamic component of the framework. Borrowing from [
is the mental action which consists in inferring from , where is an atom: an agent, believing that is true and having in its long-term memory that implies , starts believing that is true. • ∩(, ) ∩(, ) is the mental action which closes the beliefs and belief under conjunction. Namely, characterizes the mental action of deducing ∧ from and . • ⊣(, ) , where and are atoms, is the mental action that performs a simple form of “belief revision”, i.e., it removes from the belief set, in case is believed and, according to the background knowledge, ¬ is logical consequence of . • ⊢(, ) , where is an atom; by means of this mental action, an agent believing that is true (i.e., it is in the working memory) and that implies , starts believing that is true. This last action operates exclusively on the working memory without recovering anything from the background knowledge. 2.2.2. Semantics
L-INF model, including what mental and physical actions an agent can perform, what is the cost of an action and what is the budget that the agent has at its disposal, what is the degree of preference of the agent to perform each action, what is the degree of preference of the agent to use a particular resource. This choice has the advantage of keeping the complexity of the logic under control and making these aspects modular.
of the static fragment L-INF. A model is composed of two parts. A corepart and a collection of packages . More specifically: Definition 2.1.
The core part of a mode l is a tuple( , , ℛ, , ) , where • is a set of worlds (or situations); • ℛ = { }∈ Agt is a collection of equivalence relati on:s o n⊆ × ; • ∶
Agt × ⟶ 2 2 is a neighborhood function such that, fo r∈eaAcght, each , ∈ , and each ⊆
these conditions hold: (C1) if ∈ (, ) (C2) if • ∶ ⟶ 2 • ∶ ⟶ 2 do ( ) anddo ( ).
then ⊆ { ∈ ∣
} , then (, ) = (, )
; Atm is a valuation function;
{do ( ),do ( )| ∈Atm ,∈ Agt,∈ Grp} is a valuation function for formulas of the forms To simplify the notation, let ( ) denote the set { ∈ ∣ } , for ∈ . The set ( ) identifies the situations that agent considers possible at world . It is the epistemic stateof agent at . In cognitive terms, ( ) can be conceived as the set of all situations that agent can retrievferom its long-term memory and reason about. While ( ) concerns background knowledge, (, ) of all facts that agent explicitly believes at world , a fact being identified with a set of worlds. Hence, is the set if ∈ (, ) that (, )
then, the agent has the fact under the focus of its attention and believes it. We say is the explicit belief setof agent at world . Constraint (C1) imposes that agent can have explicit in its mind only facts which are compatible with its current epistemic state. Moreover, according to constraint (C2), if a world is compatible with the epistemic state of agent at world , then agent should have the same explicit beliefs at and . In other words, if two situations are equivalent as concerns background knowledge, then they cannot be distinguished through the explicit belief set. This aspect of the semantics can be extended in future work to allow agents make plausible assumptions.
The packages of a model can be thought as modular extensions of the core part. Each package is used to specify a specific feature, such as preferences, costs, executability, etc. Ideally, each package, (may) correspond to some syntactic element of the syntax of L-INF. The connection between the syntactic elements and the corresponding package will be established by a suitable component of the semantics (so be seen). The following are some possible packages. Note that we are focusing on those of interests for the purposes of this paper. Plainly, the designer of a particular MAS may decide to include only part of the following packages or even to add/model other features (also providing a suitable adaptation of the notion of truth).
Definition 2.2. Given a core mode l = ( , , ℛ, , ) executability for mental actions , the packages are: • ∶ Agt × ⟶ 2 ℒACT is an executability function of mental actions such that, f∈orAegatcahnd , ∈ , it holds that: (D1) if then(, ) = (, )
; budget and costs for mental actions (E1) if then 1(, ) = 1(, ) ; (F1) if then 1(, , ) =
1(, , ) ; executability for physical actions
(G1) if then(, ) = (, ) budget and costs for physical actions • 1 ∶ Agt × ⟶ ℕ is a budget function such that, for e∈acAhgt and , ∈ , the following holds • 1 ∶ Agt × ℒACT × ⟶ ℕ is a cost function such that, for e∈acAhgt, ∈ ℒ ACT, and , ∈ holds that: , it • ∶ , ∈
Agt × ⟶ 2 is an executability function for physical actions such that, f∈orAegatcahnd , it holds that: agents’ roles
(G2) if then (, ) = (, ) preferences on physical actions • ∶ , ∈
Agt × ⟶ 2 is an enabling function for physical actions such that, f o∈r Aeagtchand , it holds that: (E2) if then 2(, ) = 2(, ) ; (F2) if then 2(, , ) = 2(, , ) ; • 2 ∶ Agt × ⟶ Amounts is a budget function for physical action, such that, f o∈r Aeagct,hand , ∈ , it holds that: • 2 ∶ Agt × AtmA × ⟶ ∈ AtmA, and , ∈
Amounts is a cost function for physical action, such that, f o∈r Aeagct,h , it holds that: ; ; • ∶ Agt × × and , ∈
AtmA ⟶ ℕ is a preference function for physical act iosnusch that, for eac∈h Agt , it holds that: (H1) if then(, , ) = (, ,
); For each and , the functio n induces a preference ord⪯e,r on Atm , such that (, , ) ≤ (, , ′ ). ⪯, ′ if equivalence of physical actions • ∶ AtmA × ⟶ 2 is a function describing a partition o f in equivalence classes (i .e., associates each physical action with its equivalence class), such tha t∈foArgetaacnhd , ∈ , it holds that: (I1) if then( , ) = (
, ) ; • ∶ Agt × × ℒ ACT ⟶ ℒACT is a selector function for physical actions tha t,agnidve n,selects one physical actio (n, , ) from the equivalence clas s of. Namely, it holds th a(t, , ) ∈ ( , ) ∧ ∀ ′ ∈ ( , ) ′ ⪯, . For each ∈ Agt and , ∈ , it holds that: (I2) if then (, , ) = (, ,
).
that the concrete implementation, in a real MAS, the specification of some packages might depend on other packages (for example, in what follows we will describe a possible implementation of that relies on the function ).
action, has to pay the cost 1(, , ) . 1(, ) is the budget that has (in ) to perform mental actions. As mentioned, concerning physical actions, we are interested in modeling situations where performing an action may require multiple resources. Hence, the cost 2(, , ) of an action (for agent in world ) is a tuple in Amounts, while the available budget is described by 2(, ) . For an agent , the set of physical actions it can execute at is (, ) . Equivalence between physical actions is determined by function . That is, ( , ) is the set of physical actions that are equivalent to in . Roles of agents (that, as we will see, afects the capability of agents in a group to execute actions) is described through . Namely, (, ) is the set of physical actions that agent is enabled by its group to perform (recall that, at each time instant, an agent belongs to a single group). Agent’s preference on execution of physical actions is determined by the function . For an agent , and a physical action , the value of (, , ) should be intended as a degree of willingneosfsagent to execute at world . Analogously to property (C2) imposed in Def. 2.1, the constrain (D1) imposes that agent always knows which mental actions it can perform and those it cannot, but if two situations/worlds are equivalent as concerns background knowledge, then they cannot be distinguished through the executability of actions. Similar “indistinguishability’ requirements are imposed for each package by conditions (E1), (F1), (F2), (G1), (H1), (G2), (E2), (I1), and (I2).
As concerns , we assume defined (e.g., by the MAS designer) a preference relation among (equivalent)
actions, for any agent . In practice, this relation might be obtained by exploiting some specific reasoning
module. Some possibilities in this sense are described in [
The decision to integrate Blueprint Personas with the L-DINF framework stems from both theoretical and practical considerations. While Blueprint Personas provide a powerful abstraction for representing patient profiles within ASP-based scheduling systems, their utility is limited by their inherently static and declarative nature. Once instantiated, these personas remain fixed, unable to evolve or adapt in response to contextual changes. ASP solvers, though highly eficient in producing globally optimized schedules, operate in a centralized and one-shot manner. As a result, they struggle to accommodate the frequent and often unpredictable disruptions that characterize real-world healthcare environments.
The introduction of the L-DINF framework is not intended to replace ASP, but rather to augment it. L-DINF conceptualizes patients (and other agents) as cognitively capable entities endowed with beliefs, intentions, and the ability to engage in ongoing deliberation. These agents are capable of dynamically perceiving environmental changes, such as a clinic becoming unavailable, and updating their internal states accordingly. This agent-based reasoning adds a layer of adaptability and proactivity that complements ASP’s static optimization, enabling systems to respond in real time without requiring full recomputation. So, the ASP program may be scheduled for periodic execution to ensure overall optimization, whereas the L-DINF framework can be utilized to manage run-time changes.
More specifically, L-DINF enriches the scheduling process through the following capabilities: • Dynamic Adaptation: Agents revise beliefs and intentions on the fly, enabling local adjustments without re-running the entire ASP program. • Enhanced Explainability: Agent decisions are grounded in explicit reasoning chains involving beliefs, preferences, and inferred intentions. • Personalized Scheduling: Cognitive profiles support individualized, goal-directed planning beyond rule-based parameter matching. • Social Coordination: Through formal group dynamics, agents can coordinate, delegate, and share resources, capabilities that are dificult to express declaratively in ASP alone.
and computational cost in large-scale deployments, its advantages in flexibility, responsiveness, and transparency make it particularly valuable in dynamic settings. Importantly, in scenarios where scheduling constraints are stable and well-defined, ASP remains the most eficient and suitable choice.
In this section, we argue that the substitution of Blueprint Personas with L-DINF agents in appointment scheduling systems is not only feasible but also conceptually coherent. This is due to the fact that the fundamental components of a persona, such as preferences, constraints, and goals, can be structurally and semantically mapped into the epistemic constructs provided by the L-DINF framework. However, while some elements map easily, others require non-trivial adaptation due to the shift from static declarative models to autonomous cognitive agents.
• Declarative Attributes → Beliefs ( )
Example: 1 ASP: disabled(p1). 2 L-DINF: B_i(disabled(p1)). i is an agent who manages the reservations • Preferences → Preference Functions (pref _do , (, , ))
Preferences in ASP (e.g., for doctors, time slots, or clinics) are modeled as weighted rules or soft constraints. In L-DINF, preferences are formalized using pref _do and scored by a function (, , ) that quantifies the desirable action of a particular agent in the world . This allows agents to rank alternatives in a principled way, much like ASP optimizers—but grounded in agent beliefs. Example: • Constraints → Feasibility Rules (can_do )
Constraints in ASP, such as distance limits or accessibility conditions, are often specified as hard constraints. In L-DINF, these are interpreted as feasibility conditions that determine whether an agent can perform a given action. This mapping is logical and local: it preserves the original semantics while embedding it in agent-specific reasoning. Example: 1 can_do_p1(slot(c3, t1)) <--accessible(c3) and distance(c3)<20. 4.2. Components that Require Adaptation • Soft Optimization → Local Intentional Reasoning: In ASP, soft constraints are globally optimized via solvers like Clingo, producing a solution that minimizes a cost function. In L-DINF instead, agents must reason locally over their preferences and constraints to select the most suitable action. This requires restructuring the optimization logic into modular, distributed reasoning procedures. • Static Personas → Dynamic Agents: Blueprint Personas are static input structures. Once declared, they do not change during runtime. In L-DINF, agents can update beliefs based on environmental perception and modify their intentions accordingly. This demands the modeling of inference rules that govern how belief updates propagate through the agent’s decision process. • One-shot Decision Making → Ongoing Deliberation: ASP computes a single solution per run. L-DINF, by contrast, supports ongoing deliberation and dynamic replanning. This means that agents can abandon intentions if conditions change or generate new ones in response to updated knowledge. This shift requires modeling the agent’s decision lifecycle, including intend, drop, and replace operations. • No Social Context → Group Reasoning and Negotiation: Personas in ASP are individual and isolated. L-DINF allows agents to join groups, share beliefs, and coordinate actions. Implementing this requires defining group membership rules, shared knowledge bases, and collaborative preference negotiation mechanisms. While powerful, this introduces a layer of complexity not present in ASP.
Persona Element Static attributes (e.g., location) Preferences (doctor, time) Access/distance constraints Soft constraints Behavioral adaptation Multi-agent behavior
L-DINF Equivalent Beliefs ( ) pref _do , (, , ) Feasibility rules (can_do ) Local reasoning and selection Intention updates, inference Group dynamics (joinA, )
Efort to Adapt
Easy Easy
Easy Moderate
High High
L-DINF provides a more expressive and adaptive framework that retains the strengths of the original personas while extending their functionality into autonomous, intelligent scheduling agents. Synthesizing we can think of using for translation of all the modules this pseudo code: 1 forall patient P: 2 B_P(prefers_clinic = C) <-- preference(P, C) 3 pref_do_P(slot(C, T), D) <-- appointment_preference(P, C, S, E) and T in [S, E] 4 B_P(sensory_sensitive(S)) <-- sensory_preference(P, S) 5 B_P(doctor_preference(T, S, Y)) <-- doctor_preference(P, T, S, Y) 6 B_P(distance_to_clinic(C, D)) <-- distance(P, C, D)
Blueprint Persona (ASP) patient(P, …) disabled(P) preference(P, C) appointment_preference(P, C, S, E) sensory_preference(P, ``noise'') doctor_preference(P, T, S, Y) distance(P, C, D) need(P, V, N) availability(C, T) alternative(C1, T1, C2, T2) L-DINF Representation agent identity (disabled(P )) i is an agent who manages the reservations (prefers_clinic = ) (appointment_time_preference(, , )) (sensory_noise_sensitive) (doctor _preference( , , )) (distance_to_clinic(, )) intend (n_sessions( )) ∧ constraint( ) can_do (slot(, )) Cl(slot(1, 1), slot(2, 2))
Initially, Mario prefers morning appointments at Clinic C1, which is located near his home. His medical profile and preferences are encoded as a traditional ASP-style persona in section 2.1. In transitioning to an L-DINF-based representation, Mario is modeled as a cognitive agent endowed with epistemic capabilities, specifically, beliefs, preferences, and intentions,which allow for autonomous and context-aware reasoning. The initial mental state of the agent is represented through the following beliefs: 1 B_mario(prefers_clinic = c1). 2 B_mario(appointment_time_preference(c1, 0800, 1000)). 3 B_mario(sensory_noise_sensitive). 4 B_mario(doctor_preference("GP", "chronic_diseases", 10)). 5 B_mario(distance_to_clinic(c1, 12)).
1 B_mario(pref_do_mario(slot(c1, 0830), 9)).
1 B_mario(can_do_mario(slot(c1, 0830)) <-- accessible(c1) and distance(c1)< 20).
+¬ (1) . Based on an epistemic inference rule, he is able to derive a revised belief about his action feasibility: ⊢(¬ (1),¬ _ _ ((1, 0830))) . This triggers a goal revision process. Recognizing the infeasibility of his original plan, Mario searches for alternative, equivalent actions: he knows that (1, 0830) and (2, 0930) are equivalent action and his degree of willingness for the second action is 8 ( ( , , (2, 0930)) = 8. ); we also know that he prefers the second action so as to form the fact _ ((2, 0930)) . This is expressed as follows:
Initially, Mario belongs to a singleton group: 1 = { } and 1 = ∅. Anna, another patient agent, is part of a group scheduled for Clinic C2: 2 = {} and 2 = { (2), _(2, 0930), _ (2)} . Mario joins Anna’s group using the group action primitive _ (( , )) and after this operation 2 = {, }. Mario now has access to the shared group knowledge base, and he can revise his beliefs and confirm his new intention: 1 B_mario(accessible(c2)). 2 B_mario(can_do_mario(slot(c2, 0930))). 3 B_mario(do_mario(slot(c2, 0930))).
appointment scheduling systems originally based on ASP and Blueprint Personas. Rather than replacing
DINF for cognitive reasoning, agent-level deliberation, and adaptive behavior. This integration addresses the limitations of static persona models by enabling agents to revise beliefs, reformulate intentions, and coordinate with others in response to real-time contextual changes. By embedding cognitive constructs such as beliefs, preferences, and intentions into scheduling agents, the L-DINF framework introduces significant advantages in terms of dynamic adaptability, personalized decision making, and transparent reasoning. Agents modeled in L-DINF are not only responsive to environmental disruptions but also capable of proactively participating in collaborative planning through shared group knowledge and structured delegation mechanisms.
We have proposed a structured translation methodology that transforms Blueprint Personas into epistemic agent models; while this extension introduces additional modeling and computational complexity, particularly in large-scale deployments, the resulting increase in system intelligence, flexibility, and robustness demonstrates its value in realistic and dynamic healthcare environments. We have also shown that the ability to reason dynamically and deliberate continuously allows L-DINF agents to operate efectively within uncertain and evolving clinical scenarios. Future work will involve implementing this hybrid ASP + L-DINF architecture in real-world systems, validating its performance through empirical studies, and demonstrating its practical feasibility across diverse scheduling contexts.
[23] S. Costantini, A. Formisano, Augmenting weight constraints with complex preferences, in: Logical
Press, USA, 2011.