This report presents the results of a text mining project focusing on Kenilworth, a novel by Sir Walter Scott which is free to access on Project Gutenberg. The primary goal is to identify up to two key characters from the text and analyse the sentiment of the discussions surrounding these characters, namely Tressillian and Queen Elizabeth, which are then visualized via two respective plots for each character. Through a systematic approach using various natural language processing (NLP) tools, this report will demonstrate how characters are portrayed and how their narrative contexts influence the sentiment of textual mentions.

In our photogrammetry project, we focused on two different methods for digitising a wooden Matryoshka doll comparing the results of two different soft wares namely 3DF Zephyr and Polycam App: the first a scientific lab-based method, the latter a more accessible method by anyone with a smartphone. The objective was to assess the extent of information loss when using simpler equipment to digitize items in comparison with results from high-tech lab-based photogrammetry practice.

We used the 3DF Zephyr method for capturing high-quality and detailed images from various and measured angles at the Digitisation Suite at the UCL. The use of professional setting at the Suite, leveraged and optimised the reconstruction of the item producing a detailed and accurate digital representation of the Matryoshka doll.

The second method explored the Polycam App which is an accessible and user-friendly application and can be used for crowdsourcing purposes for the preservation of cultural heritage. This method focused on evaluating the loss of information that comes with generating a 3D model on the app and how it might affect the research and the long-term preservation of the item.

Our main findings present the trade-offs between a more scientific approach and mass produced more accessible results of the digitisation processes. We discuss in our group report how researchers or creators can use the two methods to tailor their approach regarding the purpose of their digitisation, financial viability, and sustainability. During our digitisation research, we discussed the capabilities and limitations of these different software solutions and how they made photogrammetry more accessible by democratizing it.

Intrigued by the challenges of capturing textiles, a notoriously difficult material in cultural heritage preservation, I brought in an unlikely test subject to the UCL Digitisation Suite: my plushie, Kaskie.

Setting up the shoot was trickier than expected. Kaskie refused to stay upright (wedges only got me so far), and I spent a good amount of time adjusting camera settings like shutter speed and lighting angles. I carried out the entire shoot using Capture One, a professional image capture and editing software, and applied a colour management workflow- including the use of a colour checker for accurate calibration.

Even with these techniques- colour correction applied to every image, professional lighting, and careful setup- it was still difficult to capture the full intricacy of the textile's texture and depth. This experiment showed just how persistent information loss can be in the digitization of fabrics, even under ideal lab conditions.

What began as a playful test with Kaskie became a meaningful exploration of the limits of digital reproduction and the complexity of preserving material presence through images.

*Note: For display purposes, the original high-resolution TIFF files have been converted to JPEG.*

I had the opportunity to work directly with UCL Special Collections, focusing on the digitization of archival material from the Jeremy Bentham Manuscripts Collection. Our group project involved digitizing two historical Bentham handlists under controlled conditions using a flat-copy digitization workflow. These included:

• Report on the Bentham MSS with Catalogue (c.1892) by Thomas Whittaker
• Catalogue of the Manuscripts of Jeremy Bentham (1962) by Alexander Taylor Milne

To ensure the integrity and accuracy of the digitization, we followed UCL's formal protocols for object handling, file naming conventions, and metadata standards, as outlined in official internal guides. All photographs were captured using Capture One in a tethered setup, with consistent use of colour correction tools, including a colour checker, to calibrate the imaging process. Each page was photographed individually under optimal lighting and camera settings not to damage the manuscript with extra light exposure.

We also discussed HDR imaging and integrated OCR techniques using Transkribus to improve text recognition, paving the way for searchable digital archives. These efforts enhance accessibility, allowing researchers to perform keyword searches and computational analysis across archival texts.

Due to archival and institutional constraints, only a small selection of sample images is shown here for demonstrative purposes.

*Note: Display images on this page were converted from high-resolution TIFF to JPEG for web presentation.*

Interactive web link to the project will be shared soon.

As part of London Data Week 2024, I co-organized a public workshop and mini-hackathon at Goodenough College London in collaboration with a member of the Coding Collective, a beginner-friendly initiative I founded to promote inclusive, community-driven tech learning. The event focused on mapping and visualizing open-source datasets from the City of London, inviting around 15 attendees to engage in hands-on exploration of civic data.

The workshop was preceded by a guest talk on cybersecurity by Dr. Vasilios Mavroudis, co-leader of the AI for Cyber Defence (AICD) Research Center at the Alan Turing Institute. A leading expert in AI and cybersecurity, Dr. Mavroudis discussed advances in AI-driven security, privacy-preserving technologies, and the future of digital defense.

We are currently preparing a follow-up event for London Data Week 2025, centered around collaborative civic mapping. Details will be announced soon.

📚 One of the attendees created a GitHub repository summarizing the event's outcomes and materials — check it out here! 🔗 https://github.com/kamalshashwat/London-Data-Week

I was invited to give a talk as part of Bülbül ve Gül, an exhibition by Moda127 Art Studio inspired by Oscar Wilde's short story The Nightingale and the Rose. In the talk, I discussed Wilde's tale alongside the process behind the exhibition — how the paintings responded to the story.

As a former member of Moda127, it was meaningful to speak about the studio's first public exhibition. Many of us came from different backgrounds beyond fine arts, and this event marked a moment of growth — seeing how the work resonated with visitors, how people engaged with the paintings, and how we'd developed as a collective over the years. It was also a way for us to open up what had mostly been a shared studio space into a public conversation.

📰 Read more about the exhibition (in Turkish) on Bant Mag and Oggito

🎥 The full talk can be watched below: