The future of urban mobility is increasingly shaped by data, driving smarter, more sustainable, and highly efficient cities. As urban centers expand, they face the pressing challenge of managing traffic congestion while ensuring accessible and environmentally friendly transportation solutions for all citizens. By integrating cutting-edge technologies, cities can now monitor traffic in real time and implement strategic interventions that improve safety, reduce emissions, and enhance overall mobility.
One of the primary catalysts behind this transformation is the ability to collect and analyze vast amounts of mobility data. This data-driven approach influences travel behaviors, optimizes transportation networks, and ensures equitable mobility benefits. A prime example of this innovation is the ACUMEN pilot project in Helsinki, where real-time data sharing is facilitated through advanced multi-layered digital twins. These digital representations of the city’s transportation ecosystem allow various systems and service providers to collaborate seamlessly, fostering a dynamic and responsive urban mobility environment.
A significant trend within this data-driven revolution is the emergence of secure, intelligent data-sharing platforms. One such example is the SEDIMARK pilot project, which has introduced a secure mobility data marketplace. The pilot is implemented as a minimum viable system and utilizes Helsinki’s mobility digital twins to create a highly adaptive model.
This marketplace is designed to provide city developers with real-time, actionable data, enabling data-driven decision-making for urban development, leading to more efficient and sustainable urban solutions—not only for mobility but also for infrastructure and condition management.
* image from Vesa Laitinen (c) Forum Virium Helsinki
As technology continues to advance, we are witnessing a transformation in how we approach problem-solving. In particular, artificial intelligence (AI) is reshaping the landscape of urban innovation, empowering individuals and communities to collaborate in creating smarter, more inclusive cities.
The Evolution of Problem-Solving Tools
In the past, challenges were tackled by specialists working in silos, relying on closed processes and limited datasets. However, with the advent of AI-powered tools, the decision-making process is becoming increasingly democratized. From machine learning algorithms and predictive analytics to intelligent automation, AI provides real-time insights that are actively shaping urban development.
One of the most notable shifts is how AI is enabling urban planners to analyze complex datasets, such as traffic patterns, infrastructure needs, and public service optimization. AI-assisted platforms are also fostering direct community engagement, with interactive dashboards and smart simulations encouraging citizens to actively participate in shaping their urban environment. Through methods like SEDIMARK data spaces and AI, we can ensure the reliability and integrity of data, which is crucial in building trust and making informed decisions. The ACUMEN pilot project uses an AI tool to promote seamless mobility, contributing significantly to the advancement of urban development.
Empowering Everyday Problem-Solvers
AI is no longer limited to data scientists or engineers. With the rise of user-friendly AI-powered tools, individuals from all walks of life are now empowered to participate in innovation. Whether it’s through AI-driven tourist assistance via chatbots, citizen-driven urban design platforms, or AI-powered tools that help communities anticipate environmental challenges, the accessibility of these tools has opened the world of innovation to a broader audience.
By breaking down barriers to information and expertise, AI fosters a collaborative environment where diverse perspectives can come together to address the challenges of urban life. This inclusivity is vital in building cities that serve the needs of all residents—not just those with specialized knowledge.
The Synergy Between AI and Human Ingenuity
While AI undoubtedly enhances problem-solving capabilities, human ingenuity remains central to the process. AI augments our ability to analyze complex data and scenarios, but creativity, ethical reasoning, and emotional intelligence continue to drive the best solutions. The future of urban innovation lies in finding harmony between AI’s computational power and human intuition, ensuring that technology becomes a tool that amplifies human agency rather than replacing it.
* image from Outi Neuvonen (c) Helsinki Partners
The document is the second deliverable of WP5: Integration, testing and evaluation of the SEDIMARK platform and reports the results of planning, execution, and evaluation of pilot trials designed to integrate the digital tools of SEDIMARK for urban and environmental management across Europe. Structured into six main sections, it provides insights into the project's objectives, methodologies, and outcomes.
The introduction establishes the purpose of the document, its relationship to other project tasks, and its overall structure. It sets the stage for understanding how the goals of the project align with broader work packages and outlines the document’s roadmap for readers.
The core of the document focuses on the refinement of trial definitions for the four distinct pilot sites, each addressing region-specific challenges.
The subsequent section delves into the integration of demonstrator systems at each pilot site. These systems are designed to interact seamlessly with the SEDIMARK toolbox. The integration of each trial is detailed, covering the systems used and their interactions, ensuring alignment with project goals.
Testing and performance assessment follow as critical components of the evaluation phase. Each pilot undergoes rigorous testing to ensure the effectiveness of the systems and their ability to meet predefined KPIs. The connection between the local systems and the SEDIMARK toolbox is emphasised as a key aspect of achieving desired outcomes.
The document concludes with a summary of the key findings and insights derived from the trials. This section highlights the challenges encountered, lessons learned, and actionable recommendations for future implementations. The project’s results serve as a foundation for advancing similar initiatives in other regions or sectors.
D5.5 deliverable can be downloaded from here.
The current document is the third deliverable of WP5 and reports the results of Task 5.2 activities regarding continuous platform integration. Building upon the foundational work presented in SEDIMARK_D5.2, this deliverable focuses on the enhancement and refinement of the platform’s capabilities through the incorporation of Minimum Viable Marketplace (MVM) and Minimum Viable Intelligence (MVI) functionalities. It emphasises the integration of decentralised data and service-sharing frameworks, advanced AI-driven tools, secure components based on Distributed Ledger Technology (DLT), APIs, and platform-wide testing. These components are designed to address medium and high-priority requirements for interoperability, trustworthiness, data quality, etc, and they were developed and described in WP3 “Distributed data quality management and interoperability” and WP4 “Secure data sharing in a decentralised Marketplace”.
The three streams will be analysed by the leaders and their contributors on the following topics:
The second release achieves the following milestones:
Following the structure of SEDIMARK_D5.2, it is important to ensure that the requirements specified in WP2 architecture and Tasks T2.1-T2.4, summarised with the SEDIMARK_D2.3, are fulfilled at all implementation phases. To serve this need, there are updated tables correlating all the medium-priority recommended requirements (M-REC) that were promised to be fulfilled for the second version of the platform, added to the existing high-priority requirements (H-REQ). The target is to monitor the status of fulfilment and in which stream they are addressed. The table covers both the functional and non-functional requirements.
The final part of the document provides an overview of the final integrated release of the SEDIMARK platform. The idea is to provide all the toolbox functionalities, where all components are in place and the system is optimised for performance purposes. Also, no hard coding is needed, and all the kinds of requirements will be fulfilled. The integration in this way will consider the timeplan for releasing the SEDIMARK integrated platform, as described in SEDIMARK_D5.1 [3]. Through continuous iterations, SEDIMARK is positioned as a robust, secure, and efficient decentralised marketplace for data and services.
D5.3 deliverable can be downloaded from here.
In the fast-evolving world of data science and AI, ensuring that workflows are reproducible, portable, and scalable is essential for success. However, many modern tools prioritize ease of use over standardization, making it difficult to share and execute workflows across different environments.
The SEDIMARK team tackled this challenge by creating a transformation methodology to convert Mage.ai workflows into Common Workflow Language (CWL) and Python-based workflows. This work aims to ensure compatibility with industry standards, improving the portability, reproducibility, and scalability of data pipelines. By integrating standardized workflows, SEDIMARK enables organizations to confidently manage their AI pipelines in secure, decentralized environments.
In the fast-evolving world of data science and AI, ensuring that workflows are reproducible, portable, and scalable is essential for success. However, many modern tools prioritize ease of use over standardization, making it difficult to share and execute workflows across different environments.
Mage.ai is a powerful tool for building data pipelines quickly and easily. However, its native workflows are not compatible with industry-standard formats like CWL, which are essential for reproducibility and portability.
CWL (Common Workflow Language) is an open standard that ensures workflows can be shared, reproduced, and executed across different platforms. It’s widely used in fields like bioinformatics and data science to standardize workflows for deployment in cloud environments, HPC clusters, and edge computing platforms.
By converting Mage.ai pipelines to CWL, organizations participating in SEDIMARK can achieve:
The SEDIMARK team developed a two-step methodology to transform Mage.ai pipelines into standardized workflows:
This transformation enables organizations to leverage SEDIMARK’s decentralized marketplace while ensuring that their data pipelines remain compatible with industry standards.
The SEDIMARK team developed a two-step methodology to transform Mage.ai pipelines into standardized workflows:
This transformation enables organizations to leverage SEDIMARK’s decentralized marketplace while ensuring that their data pipelines remain compatible with industry standards.
The SEDIMARK project aims to establish a distributed data marketplace, where organizations can securely exchange data pipelines, AI models, and other digital assets. The transformation from Mage.ai to CWL ensures that these assets are portable, reproducible, and compatible with existing standards.
For data scientists, engineers, and organizations participating in SEDIMARK, this transformation bridges the gap between intuitive pipeline design and standardized execution, enabling scalable and secure data processing.
As part of the SEDIMARK toolbox that Users will use for configuring AI and Data Processing pipelines for their use case, AI tasks such as forecasting are made readily available for inferencing on Data Assets. Time series forecasting has a wide range of applications across various fields, including financial market prediction, weather forecasting, and traffic flow prediction.
In this tutorial, we will use Python to demonstrate the basic AI workflow for time series forecasting, specifically focusing on temperature forecasting for agriculture use cases. Accurate temperature forecasting is crucial for agriculture as it helps farmers plan their activities, manage crops, and optimize yields.
The Jupyter notebook that contains the content of this tutorial can be downloaded from Github.
We need to install some toolboxes and libraries for this experiment. Therefore, please copy and use the below command in your python terminal:
pip install numpy pandas matplotlib scikit-learn torchIn this section, we generate the simulation data and apply the preprocessing.
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.preprocessing import MinMaxScaler
# Generate sample data
date_rng = pd.date_range(start='2023-01-01', end='2023-06-30', freq='D')
df = pd.DataFrame(date_rng, columns=['date'])
df['temperature'] = np.random.randint(20, 35, size=(len(date_rng)))
# Set date as index
df.set_index('date', inplace=True)
# Visualize data
df['temperature'].plot(figsize=(12, 6), title='Temperature Time Series')
plt.show()
# Normalize data
scaler = MinMaxScaler(feature_range=(0, 1))
df['temperature_scaled'] = scaler.fit_transform(df['temperature'].values.reshape(-1, 1))
# Split into training and testing sets
train_size = int(len(df) * 0.8)
train, test = df[:train_size], df[train_size:]
# Create dataset for Transformer
def create_dataset(data, time_step=1):
X, Y = [], []
for i in range(len(data) - time_step - 1):
X.append(data[i:(i + time_step), 0])
Y.append(data[i + time_step, 0])
return np.array(X), np.array(Y)
time_step = 10
X_train, y_train = create_dataset(train['temperature_scaled'].values, time_step)
X_test, y_test = create_dataset(test['temperature_scaled'].values, time_step)
# Convert to PyTorch tensors
import torch
X_train = torch.tensor(X_train.reshape(X_train.shape[0], time_step, 1), dtype=torch.float32)
y_train = torch.tensor(y_train, dtype=torch.float32)
X_test = torch.tensor(X_test.reshape(X_test.shape[0], time_step, 1), dtype=torch.float32)
y_test = torch.tensor(y_test, dtype=torch.float32)We use the toolbox and librires support provided by Pytorch to create a simple and basic Transformer model (Encoder-Decoder).
import torch.nn as nn
import torch.optim as optim
class TransformerModel(nn.Module):
def __init__(self, num_heads, d_model, num_encoder_layers, num_decoder_layers, dff):
super(TransformerModel, self).__init__()
self.encoder_layer = nn.TransformerEncoderLayer(d_model=d_model, nhead=num_heads, dim_feedforward=dff)
self.transformer_encoder = nn.TransformerEncoder(self.encoder_layer, num_layers=num_encoder_layers)
self.decoder_layer = nn.TransformerDecoderLayer(d_model=d_model, nhead=num_heads, dim_feedforward=dff)
self.transformer_decoder = nn.TransformerDecoder(self.decoder_layer, num_layers=num_decoder_layers)
self.flatten = nn.Flatten()
self.dense1 = nn.Linear(d_model * time_step, dff)
self.dense2 = nn.Linear(dff, 1)
def forward(self, src):
encoder_output = self.transformer_encoder(src)
decoder_output = self.transformer_decoder(encoder_output, encoder_output)
flatten_output = self.flatten(decoder_output)
dense_output = self.dense1(flatten_output)
output = self.dense2(dense_output)
return output
# Hyperparameters
num_heads = 2
d_model = 64
num_encoder_layers = 2
num_decoder_layers = 2
dff = 128
# Create model
model = TransformerModel(num_heads, d_model, num_encoder_layers, num_decoder_layers, dff)
criterion = nn.MSELoss()
optimizer = optim.Adam(model.parameters(), lr=0.001)
# Train model
num_epochs = 50
batch_size = 64
train_loader = torch.utils.data.DataLoader(dataset=list(zip(X_train, y_train)), batch_size=batch_size, shuffle=True)
for epoch in range(num_epochs):
model.train()
for batch_X, batch_y in train_loader:
optimizer.zero_grad()
outputs = model(batch_X)
loss = criterion(outputs.squeeze(), batch_y)
loss.backward()
optimizer.step()
print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {loss.item():.4f}')We evaluate our trained model on the created data.
import math
from sklearn.metrics import mean_squared_error
model.eval()
with torch.no_grad():
train_predict = model(X_train).squeeze().numpy()
test_predict = model(X_test).squeeze().numpy()
# Inverse transform the predictions
train_predict = scaler.inverse_transform(train_predict.reshape(-1, 1))
test_predict = scaler.inverse_transform(test_predict.reshape(-1, 1))
y_train = scaler.inverse_transform(y_train.reshape(-1, 1))
y_test = scaler.inverse_transform(y_test.reshape(-1, 1))
# Calculate RMSE
train_score = math.sqrt(mean_squared_error(y_train, train_predict))
test_score = math.sqrt(mean_squared_error(y_test, test_predict))
print(f'Train Score: {train_score} RMSE')
print(f'Test Score: {test_score} RMSE')
# Visualize predictions
plt.figure(figsize=(12, 6))
plt.plot(df['temperature'], label='Actual Data')
plt.plot(df.index[time_step:train_size], train_predict, label='Train Predict')
plt.plot(df.index[train_size+time_step+1:], test_predict, label='Test Predict')
plt.legend()
plt.show()This tutorial demonstrates how to use a basic Transformer model for time series forecasting, specifically for temperature prediction in agriculture. Accurate temperature forecasting is essential for agricultural planning and decision-making, helping farmers optimize crop management and improve yields. Through this example, readers can gain a fundamental understanding of applying Transformers to time series forecasting and further research and optimize the model for better prediction performance.
We are thrilled to share our recent presentation of SEDIMARK at the CONFORTage General Assembly Open Day!
SEDIMARK represents a significant step forward in securely integrating marketplace datasets with clinical environments, aligning closely with the European Health Data Space (EHDS) vision.
Our presentation highlighted how SEDIMARK ensures that health records remain securely within clinical domains, while offering anonymized access for external analysis. By leveraging advanced pipeline orchestration and cutting-edge integration technologies, SEDIMARK provides a secure, efficient, and interoperable platform for health data flow. This innovation is set to revolutionize decision-making and research in healthcare, enabling stakeholders to tap into new insights while maintaining stringent data security standards.
We extend our heartfelt thanks to all attendees for their enthusiastic engagement and thought-provoking questions. Your interest fuels our commitment to pioneering secure, decentralized data solutions in healthcare.
This presentation also stands as a testament to the dedication of our amazing team and collaborators. Their expertise and relentless support are the backbone of SEDIMARK's success and ongoing development.
SEDIMARK is more than a technology—it's a vision for a future where health data is managed securely, efficiently, and collaboratively. We’re eager to continue this journey, explore new partnerships, and contribute to a more innovative and secure healthcare ecosystem.
Let’s shape the future of health data management together!
SEDIMARK participated in one of the meetings organized by the ETSI cross cutting Context Information Management (CIM), providing two contributions on relevant topics regarding Context Information standardization. The following presentations were carried out by the Universidad de Cantabria team:
The meeting took place last October 9th, and all the contents and outcomes can be found in the following link.
After the thrilling session that was held at the EBDVF 2024 in Budapest on October 4th, titled "Leveraging Technologies for Data Management to Implement Data Spaces," where around 30 attendees gathered to hear representatives from six key projects—SEDIMARK, Waterverse, EnRichMyData, GREEN.DAT.AI, DATABRI-X, and STELAR—present their latest developments, we come to summarize the main aspects covered in the session and the outcomes coming out from it.
Dr. Luis Sánchez, Technical Coordinator of the SEDIMARK project, was there to present SEDIMARK’s latest developments on its presentation: “SEDIMARK: A Decentralised Marketplace for Trustworthy and Enriched Data and Services Exchange”.
Discussion topics and insights
The aforementioned projects are working to foster greater interoperability, standardization, and the seamless sharing of high-quality data across various sectors and domains. During the session, the presenters provided insights into how their respective projects are supporting the creation of Data Spaces. They highlighted a mapping between their advancements and the different Building Blocks outlined in the DSSC (Data Spaces Support Centre) Blueprint, showing how these technologies fit into the larger framework of interconnected Data Spaces. By aligning their work with the DSSC Blueprint, the projects are contributing to a more cohesive and interoperable ecosystem for data sharing.
The discussion that followed between the project experts and the audience underscored the critical need for interoperable data, which is vital for facilitating collaboration and innovation across industries. Participants emphasized the importance of establishing clear mechanisms that enable data providers and consumers to know precisely what data and services they are accessing, while also ensuring pathways for improving the quality of this data. The session also highlighted the importance of ensuring that data can flow smoothly between different systems and platforms, fostering trust and transparency in the process.
Moreover, the complementarity of the different projects' developments emerged as a key point, opening up opportunities for future collaboration. By leveraging the unique strengths of each project, the panellists and attendees recognized that a unified approach to implementing Data Spaces could accelerate progress toward a more interoperable and standardized data-sharing infrastructure across sectors.
Take-aways In conclusion, the session served as a pivotal forum for discussing the next steps toward realizing fully operational Data Spaces, emphasizing both the technological advancements being made and the ongoing need for collaboration to ensure the success of this ambitious vision.