DSSG projects 2024

Yedidim Siua Ba-Drachim

About the organization:

“Yedidim Siua Ba-drachim” is an Israeli volunteer-based organization dedicated to providing roadside assistance and support to drivers in need. Operating 24/6, its network of thousands of volunteers helps with tasks like jump-starting vehicles, changing flat tires, unlocking car doors, and addressing minor mechanical issues. The organization operates free of charge, relying on a spirit of mutual aid and community support to ensure safe and efficient travel for all. Yedidim also extends its services to non-roadside situations, offering humanitarian assistance during emergencies and crises, showcasing its commitment to fostering solidarity across Israeli society.

About the project:

In cases where the stranded driver is in need of a replacement wheel, a need arises to assess the technical parameters of the spare wheels donated for such cases. This includes the number of bolt holes, and several diameters within the wheel, which typically need to be measured manually, consuming time and energy.

The contribution:

 The project replaces the need for manual measurement by applying vision processing algorithms to a photo of the wheel, taken on the volunteer’s smartphone. Given such a photo, a pipeline of algorithms is run to determine the number of bolts, overall diameter, core bore diameter, and pitch circle diameter (PCD). The algorithm is capable of denoising the image from real-life artifacts such as bad lighting conditions, misaligned camera holding, and others.

Deliverables:

1. Algorithms for detection of wheel parameters from images
2. A cloud service to expose the algorithm to clients
3. A telegram bot for easy integration into the volunteers’ workflow
4. Final presentation

The Israel Oceanographic and Limnological Research Institute

About the organization:

The Israel Oceanographic and Limnological Research (IOLR) is a national research institution (non-profit governmental corporation) established in 1967 to generate knowledge for the sustainable use and protection of Israel’s marine, coastal and freshwater resources.

About the project:

Cyanobacteria, or blue-green algae, are key primary producers in aquatic environments, contributing to global primary production and nitrogen fixation. They support ecosystems by releasing oxygen, sequestering carbon dioxide, and serving as food for higher trophic levels. However, cyanobacterial blooms can disrupt ecosystems and produce toxins, posing health risks to humans, wildlife, and pets. These toxins can cause liver damage, gastrointestinal issues, and neurological effects, leading to health advisories and water closures. In our work, we will present a new method to predict these blooms, and in particular in the lake of Kinneret (“the sea of Galilee”).

The contribution:

An innovative method for predicting cyanobacteria without the need for chemical analysis of water samples, which often takes more than a month to obtain. By leveraging satellite image analysis, our approach considers the interconnectedness between different areas of the lake and the influence of natural conditions such as wind, water flows from streams, and rainfall. This approach not only reduces the time and effort required but also provides a more efficient and timely prediction model. Our complex analysis is framed within multiple time series algorithms to optimize prediction accuracy, achieving over 80% accuracy for forecasts up to six weeks in advance. This novel approach enhances current predictive models by incorporating spatial and environmental dynamics, offering a more holistic and accurate tool for managing cyanobacterial blooms in aquatic systems.

Deliverables:

1. Code for data cleansing, aggregation, and integration
2. Machine-learned models, calibrated and tested
3. Final Report

Daf Chadash

About the organization:

“Daf Chadash” (Fresh Start) is an Israeli organization focused on helping individuals rebuild their lives after significant personal or financial crises. By providing personalized support, counseling, and resources, the organization empowers people to regain stability and independence. It assists with job placement, financial planning, legal aid, and emotional support, tailoring its services to the unique needs of each individual. Driven by a mission to offer hope and second chances, Daf Chadash creates a pathway for its beneficiaries to achieve sustainable recovery and reintegrate into society with dignity and confidence.

About the project:

The project leverages the 2019 Insolvency and Economic Rehabilitation Law, which provides a structured legal process to help individuals settle debts and begin anew. Through legal and rehabilitative support, the project seeks to investigate critical questions about the insolvency process, such as the primary creditors, average debt amounts leading to insolvency, and the financial outcomes for all parties involved. By employing machine learning tools, the project analyzes publicly available legal records to create a data-driven understanding of insolvency trends, creditor behaviors, and debtor profiles. This empowers “Fresh Start” to advocate for systemic changes, including promoting more responsible lending practices and intervening earlier to prevent financial collapse, ultimately reducing the societal burden of insolvency.

The contribution:

The scope of this project involves processing and analyzing approximately 3,000 court rulings related to insolvency, obtained in raw formats such as scanned PDFs and DOC files. The main tasks include cleaning the data, unmasking hidden pages, and converting all documents into structured, machine-readable text. The project extracts predefined variables, such as case details, debtor demographics, financial information, and legal outcomes, despite the unstructured nature of the text. Additionally, these variables are compiled into a new, comprehensive database to enable detailed analysis. Leveraging machine learning techniques, the project automates the extraction process, ensuring high accuracy and scalability for future research.

Deliverables:

1. A GUI tool capable of ingesting the natural-language text of a court ruling, and extracting variables of importance to Daf Chadash.
2. The tool includes preliminary data analysis and visualizations capabilities, to facilitate early insights into trends and correlations.
3. Given a large number of court rulings, process each to extract the data and output the result as a single tabular database for further research.
4. Final Report
5. A short clip describing the project

The Jerusalem Development Authority

About the project:

Jerusalem is one of the most sensitive and complex cities in the world historically and culturally – being a spiritual center for the three major monotheistic religions, a city with a built and written history of over 3000 years. 

In recent decades, Jerusalem has been under enormous development pressures (the population is expected to double over less than two decades), which constitutes a substantial threat to its unique qualities. Jerusalem, therefore, might also be seen as an ‘accelerated time machine’ of sorts.

These circumstances result in a pressing question: How can the city respond to the development pressures without losing its unique qualities? Though originally formulated for the city of Jerusalem, this is an issue that is applicable to any city undergoing urban renewal.

In an attempt to provide a meaningful answer, a multidisciplinary approach needs to be employed, involving inquests and inciting public discussion regarding the question “What is my Jerusalem?” as well as a professional mapping of the urban sensitivity based on historical data, building typologies, cultural landscapes, community, and spiritual attributes. To complement these two aspects, we follow a data-driven approach utilizing computational urban morphometrics methods to cluster the city into urban archetypes and perform statistical analyses on them.

The contribution:

The project leverages the Momepy Python library to analyze urban morphometrics, focusing on the quantitative study of urban forms such as streets, plots, and buildings. By integrating Geographic Information Systems (GIS) data, including open-source resources like OpenStreetMap, the project preprocesses and standardizes geospatial data. Key contributions include tessellating urban spaces into morphological and enclosed tiles to better represent urban landscapes, calculating diverse metrics such as building areas, interbuilding distances, and street connectivity, and enabling relational analyses of spatial distributions.

Deliverables:

A graphical user interface with the following features:

1. Data Preprocessing and Metric Calculation: Users can upload building and street data in various formats. The app supports metric selection for preprocessing, calculates metrics like building heights and spatial relationships, and visualizes these metrics on an interactive map.
2. City Texture Classification: The app clusters urban data into city textures using customizable clustering methods, providing recommendations for the optimal number of clusters. Users can adjust clustering parameters, visualize classifications, and download results for further analysis.
3. Detailed Statistical Analysis: Offers statistical summaries for global datasets or specific clusters. Advanced insights include flexibility scores and variance inflation factors, aiding in data interpretation and planning applications.
4. Final Report
5. Software for free download