Liexuan Zhang's Blog

Week 1

It doesn't seem like I'm learning anything new, but I've realised that there are 4 Assessments for this module, which is making me feel a bit anxious. Also, my girlfriend asked me to make a European travel plan for reading week, but I suddenly realised that there is no reading week at the Uni of Leeds. 😡

Week 3

Things are finally getting better, the html templet is a real blessing and my site looks much better. But the process still exposed some of my current problems, for example, the inserted images may look very strange because of the proportion, I guess it has something to do with CSS, so I still need to look into it further. Oh yeah, in this week's workshop we learnt about collecting online data using an online data scraper, which was a bit easier for me as I learnt similar data collection methods and data visualisation methods in my undergraduate studies and applied them to my assessment. But the hilarious thing is that I forgot what that data visualisation tool was called.

Maybe next week I should look into adding a button under the weekly content that can link to another site so I can include some relevant screenshots and some more detail.

Week 5

I hate the flu! It's just so unpleasant.

The main difficulty I've had so far is that many of the digital scrapers mentioned on workshop won't install on my MAC, some are system incompatible while others require payment. But luckily I found a plugin called ‘Easy Scraper’ and I'm happy with it so far 😁.

Also, I found that the data collected by the data scraper seems to contain some content that I don't need, and I tried to solve this problem, but to no avail. It seems I can only manually filter and delete this useless information.

Here are my reflections on the key reading for week 5. For the dissemination of complex information, data visualisation can distil large amounts of data into a more accessible visual form, allowing audiences to grasp complex information quickly and intuitively. Furthermore, at the level of influencing perception, data visualisations can shape the way data is perceived and interpreted. They can highlight certain aspects of the data while downplaying others, thereby influencing the audience's understanding and perspective. Effective visualisations can therefore engage audiences and increase data transparency, making them valuable tools for advocacy and social justice. They can help convey important messages and mobilise support for social movements. For example, after I recently saw some statistical visualisations of data on cancer cases caused by smoking, I became frightened and determined to quit smoking. However, there may also be limitations to the information that can be conveyed by data visualisations, namely subjectivity and bias. In other words, the design choices made when creating visualisations can introduce bias as they reflect the priorities and perspectives of the creator. This can lead to misinterpretation or distorted representations of the data. And the difficulty in representing data that is complex can be significant; although visualisations are intended to simplify data, they can still be complex and may not clearly communicate the intended message to all audiences. Different audiences may interpret the same visualisation in different ways, leading to confusion or misrepresentation. Yet in their efforts to make data easy to understand, technicians may perhaps oversimplify that data, thereby ignoring important nuances and context. This oversimplification can lead to a lack of depth in the understanding of the data and its implications.

Week 7

This week's TASK confused me a lot. I visited 32 friends' Instagram homepages and used 15 of their latest posts as a source of data. And I was asked to categorise this data based on the 13 categories given and each post could not be categorised twice. This really confused me as the types of posts were really very difficult to define. After a few attempts I gave up on the categorisation job temporarily and instead started with a general overview of all the data content. I found that the attributes of each single piece of data could be classified into three types, photos with text, photos without text, and videos. After that I started defining the 12 categories(I deleted 1 category because I couldn't find anything that could be categorised as ‘Activism’)as detailed below:

1. ‘Family/partner’: photos or videos containing family members or partners fall into this category, regardless of the background of the photo. 2. ‘Outdoors’: photos or videos of outdoor scenery, which may not contain the account owner himself but can contain some passers-by, and the content theme can only be scenery. 3. ‘Work’: media content for the account owner's work scene and words, such as wearing work clothes or in the company, promotional content does not belong to this category. 4. ‘Jokes/memes ‘: no need to explain too much here, because this classification is easy to identify. 5. ‘Product/advertising: content that promotes one's own products and also includes media content that mentions other merchants or products. For example, restaurant or hotel visiting content.6. ‘Politics/news’: no explanation needed.7. ‘Music/sport/film’: no explanation needed.8. ’ Animals": any content containing any non-human animal is preferred. 9. “Friends” any scene containing a photo of the account owner and his/her friends is preferred. 10. ’Local Events": any large gathering or festival such as a concert, New Year's fair, etc. However, any scene that includes promotions, pets, friends, or any other content that is not included in this category will be given priority. However, media content containing selfies, promotions, pets, friends, family, mates, political marches, etc. do not belong to this category. 11. ‘Thoughts/reflections’: text-driven post content and must be the account owner's feelings about something, such as trolling, emotional outbursts, and reflections and summaries of a certain phase of life, but does not contain comments on political events or news content. 12. ‘Lifestyle’: selfies of the account owner alone in any scenario are preferentially categorised as this.

After summarising and analysing these sort of data, it seems that I can make a rough sense of the online identities of these 32 people. In other words, I can get a brief understanding of these 32 people's daily lives, interests and preferences, social behaviours, etc. based on these data after classification.

Week 14

Special Collections Field Trip

To be honest, this week's Special Collections Field Trip has made me feel an unprecedented challenge. Especially when I learned that these old papers, which I regarded as oracle bones in my eyes, would be related to my subsequent assessments.

Reading the Quarry Hill Unhealthy Area reference book and the Knaresborough Workhouse Daybook gave me an intimate glimpse into past living conditions and social inequalities in Yorkshire. These documents revealed not only economic hardships but also the policies and social structures that shaped people’s lives. Furthermore, the documents involving the identity information of child laborers also aroused my complex emotions. This prompted me to reflect on whether collecting such information is in line with moral standards. However, from another perspective, in order to effectively enforce the current laws and regulations, maintain the factory's rules and regulations, and even protect the rights and interests of child laborers, these pieces of information seem indispensable. This issue involves numerous factors, leaving me perplexed.

However, I must admit that language and cultural barriers do exist. First, deciphering old handwriting and terminology was difficult, as many records were written in outdated English with unfamiliar phrasing. Additionally, as someone unfamiliar with the region’s local history, I sometimes struggled to fully grasp the historical context. Another challenge was access—some records were fragile or restricted, requiring careful handling and patience.

Week 16

This week, our group worked with historical archives, extracting identifiable information and using it to craft a story.

Our first attempt was to use an OCR tool to extract text from historical documents. Unfortunately, the free OCR tool we tried struggled to recognise handwritten text in images – which, ironically, turned out to be the most crucial part of the information for us. That said, we did manage to find some legible details.

The data we uncovered were from a factory accident report. The injured worker was named Christopher Scott, a male Cloth Fuller, who was hurt on 5th May 1911 at the age of 27. The report stated that he fell to the ground from a wooden table while stacking fabric. This was the extent of the information we could extract from the archive.

However, we wanted to go further. Using historical context, social customs, and records from the same period relating to child labour and workers’ housing, we reconstructed Christopher’s life before this accident. We imagined that he might have been born in Scotland, moved with his family to Yorkshire during the Industrial Revolution, lived in back-to-back housing, and started working in a factory at the age of thirteen to support himself. This process of creating a narrative based on limited historical data made me reflect on the gaps in historical records and the cultural barriers that might lead to misinterpretations or over-extrapolations of the available information.

Another interesting aspect of this project was our use of CoPilot to generate images that complemented the story. This experience made me realise how AI-generated visuals can seem convincing to us but may be highly inaccurate in terms of historical reality. I also discovered that to produce more historically reasonable images, we needed to provide clear and precise prompts to CoPilot. Otherwise, it would generate completely unrealistic depictions – for example, if we didn’t specify the factory setting, it might portray a well-groomed Cloth Fuller in an immaculately clean, library-like textile mill, falling from an ornately crafted wooden desk. This entire process has highlighted the importance of historical accuracy and the challenges of working with both archives and AI-generated content. It’s made me more aware of how easily modern tools can distort the past – and how careful we need to be when using them.