2 Preliminaries

We assume familiarity with syntax and semantics of propositional logic and first-order logic (FO). To keep the presentation light, we develop our theory for propositional logic, but, in fact, the concepts we define apply to all logics with a model semantics. Our examples use a fragment of first-order logic.

A vocabulary $\Sigma$ is a set of symbols. A theory is a set of formulae constructed from these symbols using the logic connectives $\lnot, \land, \lor, \Rightarrow, \Leftrightarrow$ (negation, conjunction, disjunction, material implication, equivalence). A partial (resp. total) structure of $\Sigma$ is a partial (resp. total) function from $\Sigma$ to $\{{\mathbf{t}},{\mathbf{f}}\}$ , giving a Boolean value (a.k.a. interpretation) to each symbol. The value of a symbol $\sigma\in\Sigma$ in a structure S is denoted by $\sigma^{\mathit{S}}$ if it exists. In this case, we say that $\sigma$ is interpreted by $\textit{S}$ , and that the fact “ $\sigma$ has the value denoted by $\sigma^{\mathit{S}}$ ” is true in S. The set of interpreted symbols of S is denoted by $\Sigma_{{S}}$ . If S is a total structure of $\Sigma$ , we can determine the truth-value of all well-formed expressions and theories e over $\Sigma$ ; the resulting truth-value is denoted by $[\![ e ]\!]^{{S}}$ . A total structure S that makes every formula in T true is a model of T; then, T is satisfied by S.

We call a partial structure S less precise than S’ (and write ${S} \leq_p {S'}$ ) if $\Sigma_{S} \subseteq \Sigma_{{S'}}$ and all facts true in S are also true in S’, that is, if S’ assigns the same values as S to the interpreted symbols of S (it may interpret more symbols). In this case, we call S’ an expansion of S. The set of total expansions of S is denoted $\overline{S}$ . A partial structure S is consistent with T if it has a model expansion for T, that is, a total expansion that satisfies T. It entails T if all its total expansions satisfy T. We say that S and S’ are disjoint if $\Sigma_{{S}}\cap\Sigma_{{S'}}=\emptyset$ . For two disjoint structures S and S’, we define ${{S}+{S'}}$ to be the structure that maps symbols $\sigma$ interpreted by S to $\sigma^{S}$ and symbols interpreted by S’ to $\sigma^{{S'}}$ .

3 Split model expansion

We first consider the generic problem of (non-interactive) model expansion in a two-theory setting, using two vocabularies, when the environment is fully known. The first vocabulary, ${\Sigma_{\mathit{env} }}$ , contains the environmental symbols. The particular environment faced by the user is described by a total structure ${S_\mathit{env}}$ for ${\Sigma_{\mathit{env} }}$ . The second vocabulary, ${\Sigma_{\mathit{sol}}}$ , is disjoint from the first and contains the decision symbols. A solution is fully described by a total structure ${S_{\mathit{sol}}}$ for ${\Sigma_{\mathit{sol}}}$ . The union of the 2 vocabularies is denoted by ${ \Sigma_{\mathit{all}}}$ .

We assume to have a theory ${T_{\mathit{env }}}$ over vocabulary ${\Sigma_{\mathit{env} }}$ that contains the laws satisfied by any environment that could be faced by the user. Any structure ${S_\mathit{env}}$ that describes a particular environment is a model of ${T_{\mathit{env }}}$ . We also assume that the constraints on the solution to the problem faced by the user are formalized by a logic theory ${T_{\mathit{sol}}}$ in the vocabulary ${ \Sigma_{\mathit{all}}}$ . This theory fully captures all knowledge about when a solution is valid in a given environment.

A total solution for such a problem is a total structure ${S_{\mathit{sol}}}$ over ${\Sigma_{\mathit{sol}}}$ such that

\[[\![ {T_{\mathit{sol}}} ]\!]^{{{S_\mathit{env}}+{S_{\mathit{sol}}}}}={\mathbf{t}}.\]

Note that ${T_{\mathit{env }}}$ is not used in the definition of the total solution because we assume that the environment satisfies ${T_{\mathit{env }}}$ already. Having ${T_{\mathit{env }}}$ as an additional input is useful in the process of finding a solution when the environment ${S_\mathit{env}}$ is not fully known, but observable, as discussed in the next sections.

The environmental symbols we use to describe the seller and the house are propositions; the decision symbols are nullary function symbols: the type of registration (whose value is either ${\mathit{Social} }, {\mathit{Modest} }$ or ${\mathit{Other} }$ ) and the tax rate (an integer). The states of affairs are described by structures over these vocabularies:

\begin{align*} {\Sigma_{\mathit{env} }} &= \{{\mathit{SocialHousing} }, {\mathit{LicensedSeller} } , {\mathit{LowRent} } \}. \\ {\Sigma_{\mathit{sol}}} &= \{{\mathit{RegistrationType} }, {\mathit{TaxRate} }\}. \end{align*}

The theory ${T_{\mathit{env }}}$ about the possible environments states that a social housing is necessarily leased at a low rent.

\[{\mathit{SocialHousing} } \Rightarrow {\mathit{LowRent} }.\]

The legislation is described by ${T_{\mathit{sol}}}$ , stating the prerequisites for the types of registration, and the corresponding tax rate (where equality is interpreted as in first-order logic):

\begin{align*} &{\mathit{RegistrationType} } {=} {\mathit{Modest} } \Rightarrow {\mathit{LowRent} }.\\ &{\mathit{RegistrationType} } {=} {\mathit{Social} } \Rightarrow {\mathit{LicensedSeller} } \land {\mathit{SocialHousing} }.\\ &{\mathit{TaxRate} } {=} 1 \Leftrightarrow {\mathit{RegistrationType} } {=} {\mathit{Social} }. \\ &{\mathit{TaxRate} } {=} 7 \Leftrightarrow {\mathit{RegistrationType} } {=} {\mathit{Modest} }. \\ &{\mathit{TaxRate} } {=} 10 \Leftrightarrow {\mathit{RegistrationType} } {=} {\mathit{Other} }. \end{align*}

A possible sale is described by ${S_\mathit{env}} {=} \{{\mathit{SocialHousing} }{=}{\mathbf{t}}, LicencedSeller{=}{\mathbf{t}}, {\mathit{LowRent} }{=}{\mathbf{t}}\}$ . The split model expansion problem has 3 total solutions:

\begin{align*} &\{{\mathit{TaxRate} } {=} 1, {\mathit{RegistrationType} } {=} {\mathit{Social} }\}, \\ &\{{\mathit{TaxRate} } {=} 7, {\mathit{RegistrationType} } {=} {\mathit{Modest} }.\}, \text{ and}\\ &\{{\mathit{TaxRate} } {=} 10, {\mathit{RegistrationType} } {=} {\mathit{Other} }.\}. \end{align*}

An interactive assistant for this theory is available online. ^{Footnote 1}

4 Interactive model expansion

Often, a problem has many acceptable solutions in a given environment. Instead of choosing a particular solution by automated model expansion (Mitchell et al. Reference Mitchell, Ternovska, Hach and Mohebali2006), it is preferable that the automated system allow the user to explore the set of acceptable solutions. For example, a property tax legislation may allow the buyer to register the sales in different ways, as in Example 1, and the system should allow them to see the various prerequisites and consequences of each valid options, for example, on the tax rate.

We call the two interacting agents System and User. System offers User the freedom to choose values for environmental and decision symbols while verifying the validity of the solution that User constructs with respect to the 2 theories, ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ . Since a system is often designed for many users, and each user wants a solution for their own environment, we assume it is the task of User to communicate information about their environment to System. User may have additional knowledge that is unknown to System. Indeed, it is often the case that some tacit knowledge is not formalized in the theories, but well known by the user. Obviously, System cannot use this tacit knowledge; instead, User must use it in the appropriate way.

At each step of an interactive process, User has constructed two partial structures, ${S_\mathit{obs}}$ and ${S_{\mathit{dec}} }$ , assigning values to environmental and decision symbols respectively. Those values correspond to observations and to (tentative) decisions respectively.

By contrast, in prior work where there is no distinction between environmental and decision symbols, the state is described by only one partial structure.

Although System only has the information that User has provided to it, it also knows ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ ; hence, it can verify that the state is consistent with ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ :

Definition 3 A state $({S_\mathit{obs}}, {S_{\mathit{dec}} })$ is consistent (with ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ ) iff there is an expansion of the state that satisfies both theories:

\[\exists S \in \overline{{{S_\mathit{obs}}+{S_{\mathit{dec}} }}} : [\![ {T_{\mathit{env }}} \cup {T_{\mathit{sol}}} ]\!]^{S} = {\mathbf{t.}}\]

During the interactive process, decisions can be added and retracted. This leads to the following definition:

In the interactive process, User and System work together towards a solution as follows:

• The interaction starts with a pair of empty structures, ( ${S_\mathit{obs}}$ , ${S_{\mathit{dec}} }$ ). If this state is inconsistent with ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ , System tells User that there is no solution to the split model expansion problem.

• User can make ${S_\mathit{obs}}$ or ${S_{\mathit{dec}} }$ more precise as long as the new state is consistent. System prevents User from changing their decisions ${S_{\mathit{dec}} }$ to an inconsistent state because it would not bring them closer to a solution. If User needs to make ${S_\mathit{obs}}$ more precise following an observation, and if this change makes the state inconsistent, System asks User to retract a decision first so that the new observation can then be added without making the state inconsistent. To facilitate this task, System shows the list of decisions that, if retracted, would allow the entry. This list is the list of interpretations in the $\leq_p$ -minimal structure S less precise than ${S_{\mathit{dec}} }$ that, together with ${S_\mathit{obs}}$ , is inconsistent with ${T_{\mathit{env }}} \bigcup {T_{\mathit{sol}}} \bigcup O$ , where O is the observation to be added. It can be obtained using unsat_core methods (Cimatti et al. Reference Cimatti, Griggio, Sebastiani, Marques-Silva and Sakallah2007). By repeatedly making the state more precise while keeping it consistent, User eventually obtains a total solution.

• User can make ${S_\mathit{obs}}$ or ${S_{\mathit{dec}} }$ less precise, that is, retract past observations and decisions. The resulting state is necessarily consistent. This allows User to explore the complete space of solutions in their environment.

Suppose instead that User selects ${\mathit{RegistrationType} } {=} {\mathit{Social} }$ followed by ${\mathit{LowRent} } {=}{\mathbf{f}}$ . The resulting state is consistent with ${T_{\mathit{sol}}}$ in the absence of ${T_{\mathit{env }}}$ . However, the state is inconsistent with ${T_{\mathit{env }}}$ because a ${\mathit{Social} }$ registration can only be done for a social housing, which is necessarily leased at a low rent. User must backtrack on their decision ${\mathit{RegistrationType} } {=}{\mathit{Social} }$ before entering the observation that the housing is not leased at a low rent.

Note that the presence of ${T_{\mathit{env }}}$ reduces the number of consistent states, and thus shortens the search for a solution.

5 Propagation

System further assists User by informing them of the logical consequences of what is known in the current state. Such knowledge is of great value to User, and is used in most interactive assistant tools (Van Hertum et al. Reference Van Hertum, Dasseville, Janssens and Denecker2017; Falkner et al. Reference Falkner, Haselböck, Krames, Schenner, Taupe, Hotz, Aldanondo and Krebs2019). It reduces the number of decisions that User has to make, and shortens the interaction in a way analogous to how unit propagation accelerates the naive backtracking algorithm in propositional satisfiability checking. The consequences of what is known are determined by a computation called propagation (or backbone computation).

Structure S’ expands S by selecting some of the uninterpreted symbols whose value is the same in all models of T that expand S, and assigning them that value. We call S’ the optimal propagation of S with respect to T if it is the most precise T-propagation. Algorithms to compute propagation are discussed in the literature (e.g. Reference Zhang, Pu, Zhang, Ameur and QinZhang et al. (2019)).

In our interactive assistant framework, a complication arises when propagation derives facts for environmental symbols using the theory of solutions. In general, there is no guarantee that these facts are true in the environment faced by the user. To address this issue, we first propagate using ${T_{\mathit{env }}}$ only, and then using both ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ . This will further demonstrate the benefits of the split theory setting.

Because ${T_{\mathit{env }}}$ only concerns environmental symbols, the propagation of ${S_\mathit{obs}}$ by ${T_{\mathit{env }}}$ can assign new values only to environmental symbols. We denote the structure containing these new values by ${S_\mathit{obs}}^e$ . The propagation of the state by both ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ , that is, by ${T_{\mathit{all }}}$ , can assign new values to both environmental and decision symbols. We denote the structures containing these new values by ${S_\mathit{obs}}^a$ and ${S_{\mathit{dec}} }^a$ . The structures ${S_\mathit{obs}}^e$ , ${S_\mathit{obs}}^a$ and ${S_{\mathit{dec}} }^a$ are disjoint. System displays these structures differently:

• ${S_\mathit{obs}}^e$ : The new environmental facts obtained by propagation of ${S_\mathit{obs}}$ using ${T_{\mathit{env }}}$ are shown as consequences to User: they do not need to be verified by observing the environment. This is safe because the propagated facts are true in the environment of User, by the definition of ${T_{\mathit{env }}}$ .

• ${S_\mathit{obs}}^a$ : The new environmental facts obtained by propagation of the state using ${T_{\mathit{all }}}$ are shown as prerequisites of his decisions, to be verified by observing the environment. Indeed, there is no guarantee that they are true in the environment faced by User. The observation of the environment may occur at a later stage of the interactive process. If the observation does match the value obtained by propagation, User adds it to the state ${S_\mathit{obs}}$ ; the state will remain consistent. If the observation does not match, adding it would make the state inconsistent: User must first retract a decision to allow the entry, as explained in Section 4.

• ${S_{\mathit{dec}} }^a$ : The new decision facts obtained by propagation of the state using ${T_{\mathit{all }}}$ are shown as consequences to User. User cannot just change such a value, since that would cause inconsistency. If User wants to update the propagated value, they first need to retract decisions that led to the propagation.

6 Partial solutions

We now consider other forms of acceptable solutions. For example, it may not be necessary to observe the environment fully. In addition, some decisions may be unnecessary, in the sense that they can be made arbitrarily without impacting the result. Finally, the user may postpone some decisions when they have the guarantee that a solution exists.

We define definite and contingent solutions. A definite solution entails the satisfaction of ${T_{\mathit{sol}}}$ , assuming ${T_{\mathit{env }}}$ :

Definition 6 A definite solution of the split model expansion problem $\mathcal{SMX}$ is a state $({S_\mathit{obs}}, {S_{\mathit{dec}} })$ such that (i) all facts in ${S_\mathit{obs}}$ are true in ${S_\mathit{env}}$ and (ii) ${T_{\mathit{sol}}}$ is satisfied by every expansion ${S_o}$ of ${S_\mathit{obs}}$ that satisfies ${T_{\mathit{env }}}$ and by every expansion ${S_d}$ of ${S_{\mathit{dec}} }$ ,

\[ \forall{S_o} \in \overline{{S_\mathit{obs}}} : [\![ {T_{\mathit{env }}} ]\!]^{{S_o}} = {\mathbf{t}} \Rightarrow \forall{S_d} \in \overline{{S_{\mathit{dec}} }}: [\![ {T_{\mathit{sol}}} ]\!]^{{{S_d}+{S_o}}} = {\mathbf{t.}} \]

Notice that Definition 6 treats ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ very differently. This is because even though we do not know the full structure ${S_\mathit{env}}$ , we do know that it exists, that it is more precise than ${S_\mathit{obs}}$ , and that it satisfies ${T_{\mathit{env }}}$ . For this reason, we only need to consider expansions of ${S_\mathit{obs}}$ that satisfy ${T_{\mathit{env }}}$ . For such expansions, it should be the case that no matter how we make our choices, we obtain a solution to the problem at hand.

Any consistent state can be expanded into a definite solution. Indeed, by definition, every consistent state can be expanded into a total solution, which is a definite solution. In practice, definite solutions are often partial, not total, structures. As soon as User has found a definite solution, they can stop making observations and decisions. Indeed, such a solution can always be expanded to construct total solutions of the problem, by further observing the environmental and making any decision for the remaining decision symbols.

In some cases, User only needs the guarantee that a solution is reachable in their environment, by making the appropriate remaining decisions in a later stage. Since they do not know their environment fully, establishing this guarantee requires considering every environment more precise than what is known:

Definition 7 A contingent solution of the split model expansion problem $\mathcal{SMX}$ is a state ( ${S_\mathit{obs}}$ , ${S_{\mathit{dec}} }$ ) such that (i) all facts in ${S_\mathit{obs}}$ are true in ${S_\mathit{env}}$ and (ii) ${T_{\mathit{sol}}}$ is satisfied by at least one expansion ${S_d}$ of ${S_{\mathit{dec}} }$ for every expansion ${S_o}$ of ${S_\mathit{obs}}$ that satisfies ${T_{\mathit{env }}}$ , that is,

\[ \forall{S_o}\in \overline{{S_\mathit{obs}}}: ([\![ {T_{\mathit{env }}} ]\!]^{{S_o}} = {\mathbf{t}} \Rightarrow \exists{S_d} \in \overline{{S_{\mathit{dec}} }} : [\![ {T_{\mathit{sol}}} ]\!]^{{{S_d}+{S_o}}} = {\mathbf{t}}). \]

Complexity The complexity of deciding whether a state is a contingent or definite solution is much higher than checking whether it is consistent and total. In the context of propositional logic, checking for definite solution is $\Pi^p_1$ -complete (coNP-complete) while checking for contingent solution is $\Pi^p_2$ -complete. In Section “Implementation,” we propose a cheap approximate method to solve this problem.

7 Relevance

We now turn to the question of determining which observations are relevant for the search. We have shown that uninterpreted symbols in a definite solution ( ${S_\mathit{obs}}$ , ${S_{\mathit{dec}} }$ ) are irrelevant (or “do-not-care”) in the following sense: ${S_{\mathit{dec}} }$ extended with any interpretation for these symbols will satisfy ${T_{\mathit{sol}}}$ in any possible environment that expands ${S_\mathit{obs}}$ while satisfying ${T_{\mathit{env }}}$ . In this section, we extend the notion of relevance to any state.

A minimal contingent solution is similarly defined; the set is denoted by $\mathit{MinCont}(S_\mathit{obs},{S_{\mathit{dec}} })$ .

Definition 9 The symbol $\sigma$ in ${ \Sigma_{\mathit{all}}}$ is relevant in the consistent state $({S_\mathit{obs}}, {S_{\mathit{dec}} })$ if, and only if, it is interpreted in at least one $\leq_p$ -minimal definite solution more precise than the state, that is,

\begin{equation*} \exists {S} \in {MinDef}({S_\mathit{obs}}, {S_{\mathit{dec}} }): \sigma \in \Sigma_{{S}}. \end{equation*}

By our definition, all symbols interpreted in the state are relevant. As soon as User enters a value for a relevant symbol, the state is updated and the list of relevant symbols can be re-computed.

These properties allow System to alert User when a definite (resp. contingent) solution has been found. In propositional logic, the problem of computing whether a symbol $\sigma$ is relevant in some state is $\Sigma^p_3$ -complete. While computing the set of relevant symbols is computationally expensive, it can be approximated cheaply, as discussed in the next section.

8 Implementation and validation

We developed a decision support tool that embodies our proposal, called the Interactive Consultant. Given two vocabularies $({\Sigma_{\mathit{env} }}, {\Sigma_{\mathit{sol}}})$ and two theories ( ${T_{\mathit{env }}}$ and ${T_{\mathit{sol}}}$ ), the tool generates a user interface that allows the user to follow the proposed interactive process, with the benefit of computer-generated guidance. The tool supports theories in ${\mathrm{FO}(\cdot)}$ , that is, in first-order logic extended with types, aggregates, linear arithmetic over infinite domains, and (possibly inductive) definitions. ^{Footnote 2} It uses the IDP-Z3 reasoning engine for ${\mathrm{FO}(\cdot)}$ (Carbonnelle et al. Reference Carbonnelle, Vandevelde, Vennekens and Denecker2022) ^{Footnote 3} , which is based on the Z3 SMT solver (de Moura and Bjørner Reference de Moura, Bjørner, Ramakrishnan and Rehof2008). The tool and its source code are available online. ^{Footnote 4}

The user enters an observation or a decision by clicking on buttons or entering values in fields. The interactive advisor (1) updates the state, (2) propagates the state using ${T_{\mathit{env }}}$ , (3) propagates again using ${T_{\mathit{env }}} \land {T_{\mathit{sol}}}$ , and (4) determines the relevant symbols. The user interface is then updated to reflect the attributes and state of symbols: environmental vs. decision symbol, relevant, propagated. When there are no environmental symbols that are both relevant and uninterpreted, the solution is known to be a contingent solution for the environment. When there is no symbol that is both relevant and uninterpreted, the solution is known to be a definite solution for the environment, and the user can stop observing and making decisions.

Because relevance, as defined in Section 7, is expensive to compute, we approximate it using a 2-step process: (1) we compute propagation in state $({S_\mathit{obs}}, {S_{\mathit{dec}} })$ as discussed in Section 5 and create theory C consisting of the propagated literals; ^{Footnote 5} (2) we replace atoms occurring in the grounded ^{Footnote 6} version of ${T_{\mathit{all }}}$ by their value in the state (including propagated values), if any, and we simplify the formula using the laws of logic and arithmetic, to obtain $\mathop{T}\nolimits_{\mathit{all }}^*$ . Then, the set of definite solutions that extend state $({S_\mathit{obs}}, {S_{\mathit{dec}} })$ for theory $C \cup T_{\mathit{all }}^*$ is the set of such solutions for theory ${T_{\mathit{all }}}$ , that is, $C \cup T_{\mathit{all }}^*$ and ${T_{\mathit{all }}}$ are equivalent in that state. The symbols that do not occur in theory $C \cup T_{\mathit{all }}^*$ cannot be relevant: indeed, changing their value cannot change the truth-value of ${T_{\mathit{all }}}$ in the state. Thus, theory $C \cup T_{\mathit{all }}^*$ contains all the symbols that are relevant in the state; it may also contain some irrelevant symbols, depending on how thoroughly the propagation and simplification is performed, but this is acceptable in practice. Since Step 1 is computed for the reasons given in Section 5, the additional cost of computing relevance is only determined by Step 2, and can be very low and still be useful to the user.

Definitions stated in the theory require special care when determining relevance. Definitions specify a unique interpretation of a defined symbol, given the interpretation of its parameters. They take the form of an equivalence between the defined atom and a definiens, that is, a logic formula that defines it. A defined atom can be made irrelevant by replacing it by its definiens in the theory: the transformed theory will be equivalent (except for the defined symbol), and the defined atom will not occur in it anymore. Yet, defined atoms are often relevant for the user as it abstracts complexity in a useful way. Thus, we distinguish definitions from the other logical sentences in the theory. We consider a ground atom relevant if, and only if, it is important for the user to know its value, or it occurs in a simplified logical sentence, or it occurs in the simplified definition of a ground atom previously considered relevant.

To validate our approach, we have configured the decision support tool for Example 1 (available online ^{Footnote 7} ). Our tests showed that the system effectively assists the user by correctly determining which hypotheses must be verified, and which observations are relevant.

We then performed experiments on the real estate legislation adapted from Deryck et al. (Reference Deryck, Vennekens, Devriendt and Marynissen2019) (available online ^{Footnote 8} ). We simulate the user by a robot entering values in the user interface in a random order. We measured how many decisions and observations the robot has to make, using two approaches:

• using the traditional approach where the problem instance is fully known before model expansion; the user enters the observations before his decisions, and stops when he finds a total solution;

• using our approach; the user only enters relevant observations and decisions, until he finds a definite solution.

In the traditional approach, the user has to make an average of 26.4 entries to obtain a model; ^{Footnote 9} with the approach we propose, the user only needs to make 11.6 entries to obtain a definite solution, a 56% reduction (see details in Appendix). This reduction in the user’s workload is also observed in other realistic use cases (Aerts et al. Reference Aerts, Deryck and Vennekens2022).

9 Related work

The configuration problem, that is, the problem of configuring a complex system out of simpler components to satisfy customer requirements, is a special case of model expansion that has been studied, for example, by Felfernig et al. (Reference Felfernig, Hotz, Bagley and Tiihonen2014b). Several authors have discussed interactive tools to solve configuration problems (Hotz et al. Reference Hotz, Felfernig, Stumptner, Ryabokon, Bagley, Wolter, Felfernig, Hotz, Bagley and Tiihonen2014; Madsen Reference Madsen2003; Van Hertum et al. Reference Van Hertum, Dasseville, Janssens and Denecker2017; Falkner et al. Reference Falkner, Haselböck, Krames, Schenner, Taupe, Hotz, Aldanondo and Krebs2019). Our work is orthogonal to these studies. Indeed, the explicit distinction we make between environmental and decision symbols, and between the theory of the environment and the theory of acceptable solutions, can be combined with the ideas from earlier work. To illustrate this, our interactive tool supports the 8 reasoning tasks identified as useful by Van Hertum et al. (Reference Van Hertum, Dasseville, Janssens and Denecker2017).

Several approaches for controlling the relevance of symbols in configuration problems have been proposed. In these approaches, the modeler has the responsibility to write so-called activation rules that explicitly specify when a symbol becomes active (as done by Mittal and Falkenhainer (Reference Mittal, Falkenhainer, Shrobe, Dietterich and Swartout1990); Bowen and Bahler (Reference Bowen, Bahler, Dean and McKeown1991) in Dynamic CSPs), or even when an entire component becomes active (as done by Stumptner and Haselböck (1993) in Generative CSPs). Our approach does not need activation rules to derive relevance of symbols. Jansen et al. (Reference Jansen, Bogaerts, Devriendt, Janssens and Denecker2016) also propose an approach without activation rules, based on justification theory. Its implementation requires a custom solver (Deryck et al. Reference Deryck, Vennekens, Devriendt and Marynissen2019). By contrast, our implementation uses a standard SMT solver.

To facilitate decision-making in companies, the Object Management Group has published a widely accepted standard in the business process communities, called the Decision Model and Notation standard (DMN) (OMG 2021). One extension of DMN, called cDMN (Vandevelde et al. Reference Vandevelde, Aerts and Vennekens2021), further increases the expressivity of the notation. Decision models in those two notations can be translated automatically into ${\mathrm{FO}(\cdot)}$ (Vandevelde et al. Reference Vandevelde, Aerts and Vennekens2021), allowing our Interactive Consultant to support decision-making in companies. These DMN representations would be enhanced by using two vocabularies and two theories, as we propose, further highlighting the relevance of our work.

Our two-theory setting has some similarity with distributed constraint satisfaction problems, that is, problems where several agents, each with their own constraints, search for a solution that satisfies their constraints (Yokoo Reference Yokoo2012). Here, the environment, System and User may be viewed as agents, each with their own set of constraints. In our case, however, the theory of the environment is used by another agent, the interactive assistant.

The concept of relevance has been discussed in a special issue of Artificial Intelligence in 1997 (Subramanian et al. Reference Subramanian, Greiner and Pearl1997), focusing on autonomous systems with limited computational resources. Relevance has been applied to improve the performance of propositional satisfiability solvers (Jansen et al. Reference Jansen, Bogaerts, Devriendt, Janssens and Denecker2016). Instead, we apply it to reduce the workload of the user, by showing them the observations and decisions that make them progress in their search of a solution.