Knowledge Models for Spatial Planning: Ecosystem Services Awareness in the New Plan of Bari (Italy)

: The concept of ecosystem services (ES) arises as a formal outcome of historical processes of understanding and interpreting settlements as complex ecological systems. Because of a straightforward, bottom-up demand for environment enhancement, this concept increasingly occurs in discourses, in narratives, in the demands of common people, triggering a new urban environmental awareness. This is now often arising spontaneously in the protocols of participatory plan processes, especially when planning for the future of complex environments such as city areas. The present study tries to elicit reﬂections around the signiﬁcance of ES issues awareness in the case study of Bari (Italy), which is experiencing an inclusive and participatory process of construction of shared knowledge for the new master plan. Starting from an initial campaign of civic walks (CWs) along the urban neighborhoods and a subsequent semi-structured interview to the community, the paper carries out comparative analyses using problem-structuring methods (PMs), in order to evaluate and reﬂect on community behaviors and expectations about ES. Then the paper ends by emphasizing the role of structured knowledge-raising approaches, as critical activities to enhance ecosystem awareness in planning settlements as complex ecological systems.


Introduction
In the Italian season of so-called third-generation plans following earlier post-war planning experiences [1], issues of qualitative (as well as interstitially speculative) transformation of cities appear, apparently in terms of urban facilities and services.
This type of approach is generally considered as extended until the 1980s, with some medium-sized cities often cited as examples, such as Pavia, Pistoia, Arezzo. This is also the period, however, of an eruption of the environmental question in scientific debates. New reflections focus on the limits of dissipative growth especially within the residential settlements. What emerges is the need for progressively increased attention to natural resources and their regeneration cycles, especially in urban areas. Some observers even deduce from this circumstance an emerging fourth generation of spatial plans, contaminated by new increasing socio-environmental operational programs, such as Agenda 21 [2].
Certainly, a new awareness is growing around the need for closing natural cycles, to avoid problems of liveability, health, consumption of ecological resources. Settlement areas are increasingly considered, planned and managed as complex ecological systems and not as simple territories to be transformed.
The hand of public administration and policymaking can do much in this framework, in its role as a service provider to support the life and welfare of communities. By the new millennium, the new The natural capital of Bari can be referred to a number of consolidated aspects, such as (i) The sea, which is embedded in the culture and economy of the city; (ii) the nine blades, organized in a complex ecological system ( Figure 1); (iii) areas of archaeological interest and public parks located in different districts. The cultural and social capital can be represented by (iv) the historical preexistences offered in the city center by houses dating back to the Byzantine era ( Figure 2); (v) the lictor architecture developed during fascism in the peripheral center and on the waterfront ( Figure  3), and the architecture of twentieth-century rationalism in the peripheral districts (PPTR Puglia 2009) [28]. The natural capital of Bari can be referred to a number of consolidated aspects, such as (i) The sea, which is embedded in the culture and economy of the city; (ii) the nine blades, organized in a complex ecological system ( Figure 1); (iii) areas of archaeological interest and public parks located in different districts. The cultural and social capital can be represented by (iv) the historical pre-existences offered in the city center by houses dating back to the Byzantine era ( Figure 2); (v) the lictor architecture developed during fascism in the peripheral center and on the waterfront (Figure 3), and the architecture of twentieth-century rationalism in the peripheral districts (PPTR Puglia 2009) [28].
While it is easier to trace the benefits for the well-being of humans and of the ecosystems offered by the natural factors (such as, for example, the role of urban green that can regulate the climate by absorbing carbon dioxide and supporting the recycling of nutrients and the formation of soil or attract people for leisure and tourism), it is harder to associate the benefits offered by the built landscape, despite building issues are central to the public discontent with natural resource management decisions [29]. Therefore, setting up a participatory process in the identification and evaluation of ES is an increasing issue in planning agendas. Starting from resolution nr. 565/2015 in 2015 Bari Department of urban planning started a participatory process at several levels, asking citizens to participate in the drafting of the master plan of the city (PUG) [30]. The path was designed in different time steps, performed between May and November 2016, through (i) nr.30 urban front offices (UFOs) activated in the municipal area (Figure 4), (ii) nr.9 civic walks (CWs) to single out peculiar aspects and features of relevant areas (Table 2; Figures 5 and 6), and (iii) nr.5 public workshops (PW). Located in different parts of the city (Figure 4), UFOs have supported the participatory process through the distribution of surveys regarding the themes of (i) public space, (ii) landscape, and (iii) mobility; CWs aim to share new visions of the places [31]. Analysts have chosen nine CWs in order to highlight the peculiarities of each area ( Table 2); experts knowledge (institutional referents and technicians) and non-expert knowledge (citizens) have been included in order to present observations, questions, and desires. PWs aimed at expanding the dissemination of participation culture in urban policies, through the proposal of some open meetings to citizenship on relevant urban topics.  While it is easier to trace the benefits for the well-being of humans and of the ecosystems offered by the natural factors (such as, for example, the role of urban green that can regulate the climate by absorbing carbon dioxide and supporting the recycling of nutrients and the formation of soil or attract people for leisure and tourism), it is harder to associate the benefits offered by the built landscape, despite building issues are central to the public discontent with natural resource management decisions [29]. Therefore, setting up a participatory process in the identification and evaluation of ES is an increasing issue in planning agendas. Starting from resolution nr. 565/2015 in  While it is easier to trace the benefits for the well-being of humans and of the ecosystems offered by the natural factors (such as, for example, the role of urban green that can regulate the climate by absorbing carbon dioxide and supporting the recycling of nutrients and the formation of soil or attract people for leisure and tourism), it is harder to associate the benefits offered by the built landscape, despite building issues are central to the public discontent with natural resource management decisions [29]. Therefore, setting up a participatory process in the identification and evaluation of ES is an increasing issue in planning agendas. Starting from resolution nr. 565/2015 in    The public housing districts to be reconnected Stanic      The reuse and redevelopment of historical fabrics in the multiethnic city Madonnella e Libertà 5.
Archaeological protection as a natural barrier to the consumption of soil and a resource for development Torre a Mare
The historical nucleuses of public buildings and the landscape resources in the suburbs San Paolo

8.
The quality of living in the large public housing districts: public space and private space in the different neighborhood settlements Japigia 9.
The public housing districts to be reconnected Stanic     Minor historical centers and hypogea Carbonara e Ceglie 2.
Relations between the city and the sea Libertà e Marconi

3.
The places of the "gigantism" of the Quaroni Plan and the city of children Carrassi e Poggiofranco

4.
The reuse and redevelopment of historical fabrics in the multiethnic city Madonnella e Libertà 5.
Archaeological protection as a natural barrier to the consumption of soil and a resource for development Torre a Mare
The historical nucleuses of public buildings and the landscape resources in the suburbs San Paolo

8.
The quality of living in the large public housing districts: public space and private space in the different neighborhood settlements Japigia 9.
The public housing districts to be reconnected Stanic    The present study starts from a participatory process carried out by the Municipality of Bari, trying to highlight the limitation of the methodology adopted and to elicit some reflections on the weight of ES in citizens awareness. Section 4 describes the proposed methodology.

Methodology
Historically, building knowledge models for spatial planning purposes has a very recent genesis. This happened essentially because of a research area, dominant from the second half of the 1900s until around the 1990s, characterized by the qualitative and discretionary data management, often referred to as "participatory planning." The rationale behind this position can be sought mainly as a response to a strong demand for bottom-up democracy in decisions, after long traditions of top-down decision-making centralism [32] (p.27). Toward the end of this exciting season, dangerous demagogic, rhetorical, and speculative pitfalls ended up characterizing and often degrading the ethical and social effectiveness of participatory planning. The increasing awareness that has fortunately followed has been paralleled by an emerging research on new methodologies offered by information and communication technology for the management of complex and widespread data. ICT-based statistical-mathematical platforms have increasingly allowed the management of multisource and multiagent data in a quantitative or quali-quantitative way [33]. A growing scientific research has emerged regarding the construction of quantitative bottom-up knowledge management models, to support more informed, real-time decisions toward more effective spatial policy and planning. For the social management of environmental resources and components of the territory, the problem is particularly complex as it is linked to an inherently embedded social and environmental complexity. In addition to mainly qualitative methods, e.g., structured in terms of building future scenarios, ontological and relational knowledge management models have evolved in an attempt to preserve such inherent complexity while maintaining a synthesis necessary to allow its management [34][35][36][37][38]. In our paper, the context is characterized by a hybrid methodological approach for which a quantitative knowledge management model accompanies and criticizes a mainly qualitative pre-existing approach.
In particular, the present study proposes an approach based on knowledge structuring to (i) overcome the limits emerging during the participatory process adopted by the municipality, and (ii) to investigate the weight of citizens' awareness about ES through the construction of knowledge models.
Specifically, the reflection on CWs raised three critical issues: (i) Numerical predominance of considerations by expert knowledge on non-expert knowledge; (ii) the lack of structured knowledge broadly following narrative patterns; (iii) a small number of participants, never exceeding 30 units.
In the present study, in order to overcome these limits, information emerged in narrative patterns deriving from CWs has been recorded and formalized using ad-hoc structuring platforms, particularly relevant to PSMs modelling area. Specifically, a qualitative analysis of the information deriving from CWs through knowledge discovery in text (KDT) was needed to build causal loop diagrams (CLDs) and semi-structured interviews (SSIs). KDT was oriented to structure problem framework [39] and build CLDs, whereas SSIs held a dual function of validating CLDs and involving a more significant sample of citizens. Figure 7 shows the developed methodology.  Table 3. Type of participatory modelling technique.

Group Model Building (GMB)
It is based on the involvement of a group of stakeholders in order to create a conceptual model. The model building process can start from a story telling, a set of interviews or narratives and it is supported by a facilitator.

Causal cycle diagrams and similar visual tools Mediated Modelling (MM)
It is based on the collect of information by a group of stakeholders in order to create a system dynamic model. The model building process use the computer models in order to create scenarios.

System dynamic model (Stella) Companion Modelling (CM)
It is based on involving stakeholders through roleplays in order to make them aware of the variety of points of view and their consequences in terms of actions.

Participatory Simulation (PS)
It is based on involving stakeholder through mobile devices and physical activities. The rules of the games cannot be modified by the stakeholders. Every decision and every interaction are registered for further analyses Agent-based model

Shared Vision Planning (SVP)
It is based on computer simulation in order to identify alternatives and trade-offs in a manner where stakeholders without modelling experience can actively participate in the modelling process.

Scenario simulation
Collaborative Learning (CL) It is based on the idea that learning is a naturally social act that takes place through communication.
The aim is to teach and learn in groups to work together on problems, complete a task, or create a product.
Workshop, seminar, and similar tool For this work, a group model building (GMB) has been used and the model has been built by the analyst following the narratives deriving from the CWs and validated by the SSIs. The analysis led on one of the nine CWs is described below, as an explicatory example. The choice to use CLDs and SSIs arise from an analysis of existing PMT. Levels of involvement could be enhanced to strengthen the strategic knowledge on the environment, scientific literacy, and the empowerment of citizens in helping to inform and monitor policies and management efforts related to the ecosystem services [40]. Some authors [41][42][43][44] have recently proposed literature reviews on PMT (Table 3). Table 3. Type of participatory modelling technique.

Typology Description Tool
Group Model Building (GMB) It is based on the involvement of a group of stakeholders in order to create a conceptual model. The model building process can start from a story telling, a set of interviews or narratives and it is supported by a facilitator.
Causal cycle diagrams and similar visual tools It is based on the collect of information by a group of stakeholders in order to create a system dynamic model. The model building process use the computer models in order to create scenarios.
System dynamic model (Stella)

Companion Modelling (CM)
It is based on involving stakeholders through role-plays in order to make them aware of the variety of points of view and their consequences in terms of actions.

Role-playing games Participatory Simulation (PS)
It is based on involving stakeholder through mobile devices and physical activities. The rules of the games cannot be modified by the stakeholders. Every decision and every interaction are registered for further analyses.

Agent-based model Shared Vision Planning (SVP)
It is based on computer simulation in order to identify alternatives and trade-offs in a manner where stakeholders without modelling experience can actively participate in the modelling process.

Scenario simulation
Collaborative Learning (CL) It is based on the idea that learning is a naturally social act that takes place through communication. The aim is to teach and learn in groups to work together on problems, complete a task, or create a product.
Workshop, seminar, and similar tool For this work, a group model building (GMB) has been used and the model has been built by the analyst following the narratives deriving from the CWs and validated by the SSIs. The analysis led on one of the nine CWs is described below, as an explicatory example.

Participatory Modelling Exercise: Civic Walks (CWs)
Civil walks are an early and widespread instrument of citizens' participation in the field of urban policies in order to activate new forms of knowledge about the city [31]. They are typically collective walks along the streets of urban neighborhoods, developed by citizens who accompany an expert scholar describing and discussing the features and issues related to the area. Through questions and answers, a mutual interaction of knowledge is established between the participants and the expert, which is oriented to enrich the knowledge base useful for planning purposes. CWs tell the dynamics, spaces, and urban regeneration through the eyes of citizens. CWs aim to enhance the environmental aspects, the old and new forms of the urban space and economy, the quality of public space in terms of beauty and sociality to calibrate the urban redevelopment actions.
According to the literature analysis, CWs seem to ensure some important features: (i) The implementation of de-professionalization visions, i.e., not only professionals shape the future of the districts; (ii) a demystification of problems, turning territorial planning into real and concrete perspectives, away from a virtual or mediatized knowledge; (iii) the democratization of knowledge and decisions, Sustainability 2020, 12, 1516 8 of 18 as many citizens are directly involved in the process of reflection and decision, especially those that represent an interest in the future of the districts [32].
In several Italian cities CWs have been used (see the case of Bologna, Ravenna, Modena, Livorno e Valdagno).
Despite these assumptions, the analyzed CWs present a preponderance of interventions by the expert knowledge. The CW analyzed here has crossed areas coming from two planning seasons of the city of Bari. The first one comes from the urban plan drafted by Calza-Bini and Piacentini in 1954 characterized by a traditional urban design (concentric and equidistant road links, called mediane, connected by radial roads). Figure 8 taken from a Google satellite shows the average morning traffic flow of a road that does not relate to the expanding Bari of the post-war years (Figure 8(1a,1b)). the environmental aspects, the old and new forms of the urban space and economy, the quality of public space in terms of beauty and sociality to calibrate the urban redevelopment actions. According to the literature analysis, CWs seem to ensure some important features: (i) The implementation of de-professionalization visions, i.e., not only professionals shape the future of the districts; (ii) a demystification of problems, turning territorial planning into real and concrete perspectives, away from a virtual or mediatized knowledge; (iii) the democratization of knowledge and decisions, as many citizens are directly involved in the process of reflection and decision, especially those that represent an interest in the future of the districts [32].
In several Italian cities CWs have been used (see the case of Bologna, Ravenna, Modena, Livorno e Valdagno).
Despite these assumptions, the analyzed CWs present a preponderance of interventions by the expert knowledge. The CW analyzed here has crossed areas coming from two planning seasons of the city of Bari. The first one comes from the urban plan drafted by Calza-Bini and Piacentini in 1954 characterized by a traditional urban design (concentric and equidistant road links, called mediane, connected by radial roads). Figure 8 taken from a Google satellite shows the average morning traffic flow of a road that does not relate to the expanding Bari of the post-war years (Figure 8(1a,1b)).   The second one is Quaroni's urban plan (1976), characterized by the gigantism of roads and buildings in view of demographic growth and city flows (Figures 9 and 10). the environmental aspects, the old and new forms of the urban space and economy, the quality of public space in terms of beauty and sociality to calibrate the urban redevelopment actions. According to the literature analysis, CWs seem to ensure some important features: (i) The implementation of de-professionalization visions, i.e., not only professionals shape the future of the districts; (ii) a demystification of problems, turning territorial planning into real and concrete perspectives, away from a virtual or mediatized knowledge; (iii) the democratization of knowledge and decisions, as many citizens are directly involved in the process of reflection and decision, especially those that represent an interest in the future of the districts [32].
In several Italian cities CWs have been used (see the case of Bologna, Ravenna, Modena, Livorno e Valdagno).
Despite these assumptions, the analyzed CWs present a preponderance of interventions by the expert knowledge. The CW analyzed here has crossed areas coming from two planning seasons of the city of Bari. The first one comes from the urban plan drafted by Calza-Bini and Piacentini in 1954 characterized by a traditional urban design (concentric and equidistant road links, called mediane, connected by radial roads). Figure 8 taken from a Google satellite shows the average morning traffic flow of a road that does not relate to the expanding Bari of the post-war years (Figure 8(1a,1b)).

Knowledge Discovery in Text (KDT) and Causal Loop Diagram (CLD)
The analysis of the information that emerged from the conversations during CWs was carried out through the knowledge discovery in text (KDT) approach [45]. The selection of this approach depends on its offering the possibility, using automated intelligent systems, to extract knowledge from unstructured texts. Specifically, through KDT it is possible to extrapolate contents from an unstructured text, despite the lexical difficulties inherent in a conversation [46].
The application to the present case study took place with the use of Rapid Miner studio software [47] and is divided into two phases: the first one, in which the text is "cleaned up," and the second one in which the analysis of contents and the construction of cause-effect links are carried out.
The first "cleaning" phase follows the following steps: (i) Tokenization, able to select the main words (tokens) included in a document according to a frequency logic; (ii) the identification of "Stopwords" in order to delete all the irrelevant words listed in the dictionaries provided by the system (deriving from the Italian dictionary); (iii) stemming, able to reduce the number of words that share the same root as a given token. The pieces of information emerging from the cleaning process are grouped into a correlation matrix (Figure 12) representing the second phase.

Knowledge Discovery in Text (KDT) and Causal Loop Diagram (CLD)
The analysis of the information that emerged from the conversations during CWs was carried out through the knowledge discovery in text (KDT) approach [45]. The selection of this approach depends on its offering the possibility, using automated intelligent systems, to extract knowledge from unstructured texts. Specifically, through KDT it is possible to extrapolate contents from an unstructured text, despite the lexical difficulties inherent in a conversation [46].
The application to the present case study took place with the use of Rapid Miner studio software [47] and is divided into two phases: the first one, in which the text is "cleaned up," and the second one in which the analysis of contents and the construction of cause-effect links are carried out.
The first "cleaning" phase follows the following steps: (i) Tokenization, able to select the main words (tokens) included in a document according to a frequency logic; (ii) the identification of "Stopwords" in order to delete all the irrelevant words listed in the dictionaries provided by the system (deriving from the Italian dictionary); (iii) stemming, able to reduce the number of words that share the same root as a given token. The pieces of information emerging from the cleaning process are grouped into a correlation matrix (Figure 12) representing the second phase.

Knowledge Discovery in Text (KDT) and Causal Loop Diagram (CLD)
The analysis of the information that emerged from the conversations during CWs was carried out through the knowledge discovery in text (KDT) approach [45]. The selection of this approach depends on its offering the possibility, using automated intelligent systems, to extract knowledge from unstructured texts. Specifically, through KDT it is possible to extrapolate contents from an unstructured text, despite the lexical difficulties inherent in a conversation [46].
The application to the present case study took place with the use of Rapid Miner studio software [47] and is divided into two phases: the first one, in which the text is "cleaned up," and the second one in which the analysis of contents and the construction of cause-effect links are carried out.
The first "cleaning" phase follows the following steps: (i) Tokenization, able to select the main words (tokens) included in a document according to a frequency logic; (ii) the identification of "Stopwords" in order to delete all the irrelevant words listed in the dictionaries provided by the system (deriving from the Italian dictionary); (iii) stemming, able to reduce the number of words that share the same root as a given token. The pieces of information emerging from the cleaning process are grouped into a correlation matrix (Figure 12) representing the second phase. Correlation (1) is a statistical technique that shows the extent to which the pairs of variables X and Y are related: where, X and Y are the frequency of the single words with the relative means and letter S indicate the standard deviation. Correlation is expressed by a number between -1 and 1 that measures the degree of association between the two words. A positive value indicates the presence of a positive association. It is obtained when high values of Y tend to be associated with high values of X and low values of X tend to be associated with low values of Y. An inverse association, on the other hand, is expressed by a negative value. That is, high values of X tend to be associated with the minimum of the values of Y and vice versa. These relations refer to the instances of the text, thus allowing the creation of a structured cause-effect relationship [48].
The analysis just described was fundamental in order to reduce the ambiguity of interpretation deriving from the reading of a textual document. Also, it allowed us to reconstruct the causal links between the variables in a more structured way and overcoming the limit of the necessary presence of the actors involved, as required by group model building technique (GMB) theory.
CLDs are a formal modelling tool of GMB. It is a method used mostly in business applications but also for natural resource management [49]. It consists of the involvement of a group of agents, stakeholders, in one or more sessions to build the conceptual model or problem situation represented by CLDs and similar visual tools, starting from a history or even from interviews, facts, and narratives [50]. The facilitator helps the group in the model construction, remaining generally neutral with respect to the content. CLDs is symbolized by variables and links with polarity representing the effect of one variable on another [51].
The elaboration process described above took place by dividing the text into three sections, relating to the three Places of reference ( Figure 11).
As a matter of simplicity, the present section shows the methodology applied to only Place 1. Correlation (1) is a statistical technique that shows the extent to which the pairs of variables X and Y are related: where, X and Y are the frequency of the single words with the relative means and letter S indicate the standard deviation. Correlation is expressed by a number between -1 and 1 that measures the degree of association between the two words. A positive value indicates the presence of a positive association. It is obtained when high values of Y tend to be associated with high values of X and low values of X tend to be associated with low values of Y. An inverse association, on the other hand, is expressed by a negative value. That is, high values of X tend to be associated with the minimum of the values of Y and vice versa. These relations refer to the instances of the text, thus allowing the creation of a structured cause-effect relationship [48].
The analysis just described was fundamental in order to reduce the ambiguity of interpretation deriving from the reading of a textual document. Also, it allowed us to reconstruct the causal links between the variables in a more structured way and overcoming the limit of the necessary presence of the actors involved, as required by group model building technique (GMB) theory.
CLDs are a formal modelling tool of GMB. It is a method used mostly in business applications but also for natural resource management [49]. It consists of the involvement of a group of agents, stakeholders, in one or more sessions to build the conceptual model or problem situation represented by CLDs and similar visual tools, starting from a history or even from interviews, facts, and narratives [50]. The facilitator helps the group in the model construction, remaining generally neutral with respect to the content. CLDs is symbolized by variables and links with polarity representing the effect of one variable on another [51].
The elaboration process described above took place by dividing the text into three sections, relating to the three Places of reference ( Figure 11).
As a matter of simplicity, the present section shows the methodology applied to only Place 1.
The KDT process filtered 826 words and selected the most common word pairs (29 relationships between variables with weight included, for negative values between −1 and −0.5 and for positive values between 0.5 and 1). The relationships have been represented using a Force-Directed Graph Drawing Algorithm (FDGA) (Figure 13) [52].
The FDGA is represented by nodes and links between nodes. The size of the node is associated with the characteristic of each of them; the distance depends on the repulsive or attractive force between nodes. The FDGA allows to build the framework of CLDs representing the knowledge model of citizen involved. The KDT process filtered 826 words and selected the most common word pairs (29 relationships between variables with weight included, for negative values between −1 and −0.5 and for positive values between 0.5 and 1). The relationships have been represented using a Force-Directed Graph Drawing Algorithm (FDGA) (Figure 13) [52] .
The FDGA is represented by nodes and links between nodes. The size of the node is associated with the characteristic of each of them; the distance depends on the repulsive or attractive force between nodes. The FDGA allows to build the framework of CLDs representing the knowledge model of citizen involved.  Specifically, on one hand, the road section is claimed to be unable to meet contemporary mobility demands. On the other hand, the problem of lacking green public spaces is due to intensive buildings and possibly worsened by the misappropriation of the few remaining areas by some private owners. A re-building of the traffic plan for the management of vehicular flows on the one hand, and the supervision by the public administration on the other, are the solutions proposed by the expert knowledge in response to the issues raised.
The second CLDs referring to Place 2 and 3 was carried out using the same procedure ( Figure  15). The KDT process filtered 826 words and selected the most common word pairs (29 relationships between variables with weight included, for negative values between −1 and −0.5 and for positive values between 0.5 and 1). The relationships have been represented using a Force-Directed Graph Drawing Algorithm (FDGA) (Figure 13) [52] .
The FDGA is represented by nodes and links between nodes. The size of the node is associated with the characteristic of each of them; the distance depends on the repulsive or attractive force between nodes. The FDGA allows to build the framework of CLDs representing the knowledge model of citizen involved.  Specifically, on one hand, the road section is claimed to be unable to meet contemporary mobility demands. On the other hand, the problem of lacking green public spaces is due to intensive buildings and possibly worsened by the misappropriation of the few remaining areas by some private owners. A re-building of the traffic plan for the management of vehicular flows on the one hand, and the supervision by the public administration on the other, are the solutions proposed by the expert knowledge in response to the issues raised.
The second CLDs referring to Place 2 and 3 was carried out using the same procedure ( Figure  15). Specifically, on one hand, the road section is claimed to be unable to meet contemporary mobility demands. On the other hand, the problem of lacking green public spaces is due to intensive buildings and possibly worsened by the misappropriation of the few remaining areas by some private owners. A re-building of the traffic plan for the management of vehicular flows on the one hand, and the supervision by the public administration on the other, are the solutions proposed by the expert knowledge in response to the issues raised.
The second CLDs referring to Place 2 and 3 was carried out using the same procedure ( Figure 15). Sustainability 2019, 11, x FOR PEER REVIEW 12 of 18 Figure 15. CLD referring to Place 3.
The CLDs show the causes and the effects that these variables entail. Following the same procedure, the other map has been built, in which the CLDs of Place 2 and 3 have been aggregated referring to the same issues. The use of land, deriving from the reduction of some road sections first conceived as urban highway and never completed, was a central theme referring to the Place 2 and 3. Specifically, two suggestions have been proposed: (i) Urban gardens for community along the roads and (ii) the reorganization with partial pedestrianization and bicycle path of the street to reduce the speed of traffic. The need to expand public spaces, by redeveloping the underutilized areas, was claimed in a different part of the district.
In the next section the validation of CLDs is explained through SSIs.

Problem Structuring Method: Semi-Structured Interviews
SSIs are commonly used in policy research and are applicable to many research questions [53]. They combine some structured, formalized questions with some unstructured exploration. They are useful when dealing with complex systems, owing to the use of spontaneous approaches able to better explore, understand, clarify answers to questions [54]. In this context, SSIs have been carried out and submitted to citizens, retracing the same places of the CWs, also aiming at checking the relevance and consistency of issue previously raised by expert knowledge.
SSIs have been structured in three sections: (i) citizen profiles, (ii) mobility issues, and (iii) public space issues. Citizens could express own preferences on a Likert (1932) 1-5 scale of agreement and to insert a free contribution on the actions to be addressed on the issues raised (figure 16). A sample of 88 citizens, divided into 53 women and 35 men, aged between 35 and 50 years, were interviewed. The CLDs show the causes and the effects that these variables entail. Following the same procedure, the other map has been built, in which the CLDs of Place 2 and 3 have been aggregated referring to the same issues. The use of land, deriving from the reduction of some road sections first conceived as urban highway and never completed, was a central theme referring to the Place 2 and 3. Specifically, two suggestions have been proposed: (i) Urban gardens for community along the roads and (ii) the reorganization with partial pedestrianization and bicycle path of the street to reduce the speed of traffic. The need to expand public spaces, by redeveloping the underutilized areas, was claimed in a different part of the district.
In the next section the validation of CLDs is explained through SSIs.

Problem Structuring Method: Semi-Structured Interviews
SSIs are commonly used in policy research and are applicable to many research questions [53]. They combine some structured, formalized questions with some unstructured exploration. They are useful when dealing with complex systems, owing to the use of spontaneous approaches able to better explore, understand, clarify answers to questions [54]. In this context, SSIs have been carried out and submitted to citizens, retracing the same places of the CWs, also aiming at checking the relevance and consistency of issue previously raised by expert knowledge.
SSIs have been structured in three sections: (i) citizen profiles, (ii) mobility issues, and (iii) public space issues. Citizens could express own preferences on a Likert (1932) 1-5 scale of agreement and to insert a free contribution on the actions to be addressed on the issues raised ( Figure 16

Results and Discussion
A qualitative analysis of the information deriving from CWs through KDT allowed us to build the CLDs in order to structure expert knowledge and SSI in order to validate expert knowledge and at the same time to structure non-expert knowledge.
The results emerging from the application of the methodology are described below. CLDs building tried to overcome the limit relating the unstructured approach emerged during CWs. The information thus emerging was subsequently connected to ES (EPA) classes, so making it possible to draw out considerations regarding the issues related to ES, by observing Table 4.

Results and Discussion
A qualitative analysis of the information deriving from CWs through KDT allowed us to build the CLDs in order to structure expert knowledge and SSI in order to validate expert knowledge and at the same time to structure non-expert knowledge.
The results emerging from the application of the methodology are described below. CLDs building tried to overcome the limit relating the unstructured approach emerged during CWs. The information thus emerging was subsequently connected to ES (EPA) classes, so making it possible to draw out considerations regarding the issues related to ES, by observing Table 4.
It can be noted that the most common drivers of change are the land use and policy. The benefits related to well-being and public health are connected through cause-effect relationships.
The "traffic plan re-building" variable emerged in all three places. Referring to Place 1, it was suggested to act on the traffic flow, through a study of vehicular flows, not being able to physically modify the undersized road section. Referring to Place 2 and 3, the construction of cycle paths, pedestrian route and urban gardens was suggested. The latter seems to meet a dual function of reducing the road section and (consequently) vehicle speed, while promoting sustainable mobility and equipping the district with urban gardens. In terms of benefits, these actions induce an improvement in the well-being of citizenship owing to the presence of areas for leisure, a decrease in vehicular traffic with more safety for pedestrians, an increase in health and clean air-related benefits.
The above statements have been submitted to citizens' opinion and degree of validation through SSIs. On the one hand, this allowed a general validation by the citizens on the issues emerged from expert knowledge, thus somehow balancing the preponderance of interventions by expert knowledge. On the other hand, it helped to bring out new issues such as waste management, the inclusion of public lighting, and the planting of new plant species. Variables have been relocated to relevant ES categories (Table 5).  The issues emerged, which are added to those already known to be derived from expert knowledge are: the planting of new tree species in order to reduce the problems linked to allergies that characterize children residing in Alcide De Gasperi street (Place 1); the strengthening of public lighting at Place 2 in order to increase pedestrian safety and finally, at Place 3, the need of improving the waste management system to guarantee adequate hygienic conditions of spaces and healthiness of air. The validation of CLDs took place through the analysis of semi-structured interviews. The preference of citizens in relation to questions have been analyzed. For the sake of simplicity one example is shown "Corso Alcide De Gasperi is undersized" (Figure 17a).
A total of 48% of citizens involved expressed an agreement at Likert scale 4, whereas 34% of citizens agreed at grade 5 ( Figure 17a). Most citizens involved acknowledges that Alcide De Gasperi street is undersized. It is possible to summarize the results obtained from the questionnaire through a histogram in which the abscissas represent the questions, and the ordinates represent the average of citizens' preference for each question (Figure 17b). A total of 48% of citizens involved expressed an agreement at Likert scale 4, whereas 34% of citizens agreed at grade 5 (figure 17a). Most citizens involved acknowledges that Alcide De Gasperi street is undersized. It is possible to summarize the results obtained from the questionnaire through a histogram in which the abscissas represent the questions, and the ordinates represent the average of citizens' preference for each question (Figure 17b).

Conclusion
The application of knowledge structuring models through PSMs aims to challenge some limits of PMT and to investigate the level of citizens' knowledge and then, the awareness about ES.
The study has brought about some general considerations, that can be synthesized as follows. First, CWs seem to be not completely able to lay out, analyze, and understand issues and problem situations emerging along walking discourses. An integration offered by other methods, such as extended SSIs, seems to be effectively integrative of the knowledge building process, being also possible to involve a greater number of people.
Second, the concept of ES seems to be now somehow inherent in the culture of citizens. In fact, virtually every action that emerged from CWs and SSIs can be assigned to a category of ES. However, some limitations still appear, such as: (i) Emerging ESs consider only the natural resource Land; (ii) the drivers used are only Policy and Land use and the benefits arising are only related to Well-being and Public Health.
Interestingly, CLDs seem to usefully integrate future-modelling activities, such as scenariobuilding models. For example, they seem to be useful to investigate the implications of citizen potential decisions on areas, as well as to facilitate citizens' knowledge about ES and, more broadly, to support the construction of collective futures. In this perspective, more work will be devoted to check such issues on different case studies.
Collective futures show up here as being certainly based on expert knowledge, but also largely non-expert, common, bottom-level knowledge. As a matter of facts, for the first time non-expert knowledge is considered in a computationally structured way and not only in qualitative sense, through the search and identification of knowledge database management models.
Actually, we must say that the results of the present work do parallel a constant evolution of attempts to structure informal data [55,56]. Stimulated by a well-known urgency of the environmental problem both at the global and at the local level, spatial decision-making and planning processes tend to confront complexity, rather than reducing or even denying it as occurred in the past [57].
Yet for this this knowledge-inclusive confrontation to be operational, intelligent and operationally refined instruments are needed. This represents a clear limitation of the hybrid

Conclusions
The application of knowledge structuring models through PSMs aims to challenge some limits of PMT and to investigate the level of citizens' knowledge and then, the awareness about ES.
The study has brought about some general considerations, that can be synthesized as follows.
First, CWs seem to be not completely able to lay out, analyze, and understand issues and problem situations emerging along walking discourses. An integration offered by other methods, such as extended SSIs, seems to be effectively integrative of the knowledge building process, being also possible to involve a greater number of people.
Second, the concept of ES seems to be now somehow inherent in the culture of citizens. In fact, virtually every action that emerged from CWs and SSIs can be assigned to a category of ES. However, some limitations still appear, such as: (i) Emerging ESs consider only the natural resource Land; (ii) the drivers used are only Policy and Land use and the benefits arising are only related to Well-being and Public Health.
Interestingly, CLDs seem to usefully integrate future-modelling activities, such as scenario-building models. For example, they seem to be useful to investigate the implications of citizen potential decisions on areas, as well as to facilitate citizens' knowledge about ES and, more broadly, to support the construction of collective futures. In this perspective, more work will be devoted to check such issues on different case studies.
Collective futures show up here as being certainly based on expert knowledge, but also largely non-expert, common, bottom-level knowledge. As a matter of facts, for the first time non-expert knowledge is considered in a computationally structured way and not only in qualitative sense, through the search and identification of knowledge database management models.
Actually, we must say that the results of the present work do parallel a constant evolution of attempts to structure informal data [55,56]. Stimulated by a well-known urgency of the environmental problem both at the global and at the local level, spatial decision-making and planning processes tend to confront complexity, rather than reducing or even denying it as occurred in the past [57].
Yet for this this knowledge-inclusive confrontation to be operational, intelligent and operationally refined instruments are needed. This represents a clear limitation of the hybrid approach used here, which represents just an explorative effort to evaluate an applied qualitative methodology, rather than propose an original one. This is in fact a path with a large experimental and area-based component, able to provide the details and the experimental assortment that guarantee the usefulness, quality, and effectiveness of the path itself. But the availability and/or construction of these instruments is still far from being in step with theoretical approaches and methodologies and requires continuous and reiterated efforts.
In many ways this recalls a famous scientific approach of Simonian memory, whose trial-and-error steps allow just the necessary clarification and operational fine-tuning to problem management [58]. It is therefore a still wide and interesting research path in perspective to which our next research efforts will be oriented as follow-up activities.
Author Contributions: The paper is the result of a joint research work carried out by authors together. Nonetheless, D.C. wrote Sections 1 and 6, S.S. wrote Sections 2, 3 and 5 whereas P.B. wrote Section 4. All authors have read and agree to the published version of the manuscript.
Funding: This research received no external funding.