Conference Agenda

The Sustainable Development Goals (SDGs) define a global framework to address social, environmental, and economic priorities through a standardized set of 232 indicators. These indicators are designed to guide policies, track progress, and inform decision-making at all levels of governance. However, while their definitions are globally agreed upon, their implementation often requires local adaptation and disaggregated insights especially in urban contexts where spatial inequalities are most visible.

To operationalize SDG indicators locally, stakeholders increasingly turn to open data, a rich and accessible resource covering demographics, mobility, environment, and infrastructure. Yet, open datasets are fragmented, heterogeneous in structure and semantics, and often lack metadata. This makes integration and computation difficult, especially when aligning real-world data with the formal definitions of SDG indicators.

We present SDG-KG, a graph-based framework that enables query-driven, explainable, and reproducible computation of SDG indicators from open data. The system combines Large Language Models (LLMs) and a dual knowledge graph architecture to bridge the gap between global indicator semantics and fragmented local data.

SDG-KG integrates two graphs:
(1) the SDG Graph, which encodes the structure and computation rules of each indicator (goals, targets, concepts, attributes); and
(2) the Metadata Graph, which represents open datasets in terms of schema, spatial/temporal coverage, and quality metrics.

Users define queries over a given territory and time range. SDG-KG applies LLM-assisted schema alignment to map dataset attributes and values (e.g., "bus" → "low-capacity transport") to SDG concepts, even across languages or inconsistent formats. It uses a Trust Score Metric (TSM) to rank datasets based on alignment with query constraints and data quality. An execution plan is then generated, translated into Cypher and Python code, and run over the selected sources.

We evaluate SDG-KG on six countries and two indicators: 11.2.1 (access to public transport) and 11.6.2 (PM2.5 pollution). SDG-KG supports interactive exploration of results, provenance tracing, and adaptation of indicators to the urban scale. In Paris, for instance, we show that computed values for 11.2.1 capture accessibility improvements linked to new transport developments.

By combining open data, knowledge graphs, and large language models, SDG-KG contributes to the emergence of semantic digital twins for sustainability. It empowers planners, researchers, and local actors to compute interpretable and actionable indicators aligned with SDG standards.

In summary, SDG-KG makes global goals locally measurable through transparent and AI-assisted indicator computation based on open data.

Urban Digital Twins (UDT) are increasingly positioned as transformative tools for urban planning, design, and management cities. Enabled by the Internet of Things (IoT), UDTs allow real-time monitoring and data-driven decision-making to support long-term strategic planning (Batty, 2018; Deng et al., 2021; Nochta et al., 2021). However, their integration into frameworks addressing liveability remains limited, with most implementations focusing on operational efficiency and infrastructure performance rather than social aspect (Yossef Ravid and Aharon-Gutman, 2022). Currently, only a small portion of the processes that influence how the social aspects of the city operate are abstracted by city digital twins (Batty, 2018). To bridge this gap, Digital twins should be built on purpose and toward a specific goal that be design and operated to extract trends of certain essential aspects of the city system that enhance liveability (Batty, 2019; Caldarelli et al., 2023). however, In Saudi Arabia, national efforts aim to establish digital twin platforms across five cities, these initiatives aim to support the nation's transformation efforts and shape the future of Saudi cities. including Mecca, while also planning to host over 30 million pilgrims annually by 2030 (CEDA, 2016). These overlapping ambitions raise critical concerns about Mecca’s liveability particularly under conditions of extreme temporal fluctuation highlighting what Rittel and Webber (1973) describe as a “wicked problem”.For this reason, the primary concentration of the study will be on whether the smart city/digital twins strategic policies effectively address the specific liveability concerns identified or if the initiatives prioritise different aspects of urban challenges within the city to ensure that digital twin initiatives align with the needs and aspirations of the residents of Mecca city. The study adopts a mixed-methods deductive research strategy, combining 450 household surveys using a two-stage selection process across 63 districts with semi-structured interviews involving 15 government officials' decision-making from various entities. The qualitative strand focuses on interpreting strategic policy pathways, while the quantitative component uses a five-point Likert scale (1–5) to capture residents’ perceptions of liveability. Descriptive statistics are employed to summarise key variables and provide an overview of how different aspects of liveability such as access to services, safety, and amenities—are perceived across the city. Preliminary findings suggest that current policies primarily focus on managing seasonal crowding during Hajj, while long-term issues such as housing, mobility, and environmental quality receive limited attention. There is also a noticeable lack of coordination between national and local strategies, and a clear disconnect between digital twin initiatives and the everyday needs of residents. Policy analysis reveals fragmented governance, weak cross-scale integration, and minimal alignment between smart city /digital twin strategies and broader sustainability goals. Furthermore, liveability concerns are often overshadowed by economic and technological priorities. This study initially identifies theoretical and practical gaps in the integration of smart city and digital twin policies within liveability frameworks. It calls for more adaptive, people-centred approaches that prioritise experiences and aspirations of urban residents. If strategically reoriented, urban digital twins have the potential to become powerful tools to inclusive, resilient, and sustainable cities.

In the Netherlands, the integrated Environment and Planning Act entered into effect in 2024. It is an all-in-one law on the physical environment providing both legal security and direct spatial planning. Local governments are required to submit and manage permit applications, notifications, and spatial planning information through the Environment and Planning Portal, to ensure compliance with the Act. In this way, they provide transparent access to rules and procedures for citizens and businesses. Furthermore, trough the Portal, many governments publish their so-called Soft Plans. They are non-biding policy documents which express local governments’ long-term visions and ambitions. They can be seen as consultation (i.e. participatory) tools which support a dialog between different initiators and governments, guiding future spatial development and planning decisions. Through these plans, parties can check whether new proposals fit within local strategic visions, area agendas or development frameworks.

The national spatial planning and decision-making process in the country is highly decentralized, involving multiple layers of government (i.e. national, provincial, municipal and waterboards) with complex regulations and coordination across local planning and environmental strategies. The Act strongly advocates for integration, through a policy track specifically dedicated to environmental visions, and local governments are expected to execute their own vision's assessment and alignment to the national and provincial plans. The local environmental policy is then implemented by local Uitvoeringsagenda’s Klimaatadaptatie, which are specific plans for adapting of infrastructure, nature, agriculture and the built environment to climate risks.

This complexity sometimes challenges the integral alignment of local policies in a wider context. For example, whether nature preservation objectives will be achieved depends not only on strengthening nature policy itself, but other related policies, such as sustainable agriculture targets to reflect the "critical values for nature".

With the increasing risks associated to the changing climate on public wellbeing and the environment happening sooner than expected, there is a pressing need for governmental efforts on adaptation of local living environments to be more efficient across administrative boundaries and sectors. Therefore, the need to create automated tool that could help connecting and coordinating local and regional strategies to better align across sectoral borders is increasing.

To evaluate the spatial implications of plans submitted under the abovementioned Act against some of the associated climate-related risks (e.g. health, water security, agriculture, drinking water availability etc.) defined by PBL The Netherlands Environmental Assessment Agency, a dedicated workflow has been developed to collect, integrate and analyse relevant qualitative data into spatially explicit model on climate risks and adaptation. Textual documents and geospatial data files under the integrated Act are gathered from the accompanying online platform using APIs. After processing, their content was further analysed using decentralized open-source Large Language model. The LLM by means of question-answer workflow, extracted, synthesized, and reasoned over larger volume of various spatial plans.

Through a use case, we show how unstructured or semi-structured data from non-binding future strategic visions, next to expert knowledge, could be used to synthesize complex multi-source information and produce insight for spatial scenario-driven exploration models.