 Hello everyone, our research is about the gender symmetries in the depiction of historical figures based on a comparison of bad Historia, Ecclesiastica and Wikipedia. Let's have a short introduction of best work Historia, Ecclesiastica, Gentes Anglophone, or the Ecclesiastical History of the English people. It is a very seminal work of early medieval history written Latin in the 8th century. This work was composed of five books in total and it documented the events and key individuals involved in the spread of questionality in England. As a product of its time, it has some unsurprising gender bias. For example, he tends to focus on male figures and often downplayed or ignored the contributions of women. I think everybody here should already know Wikipedia quite well, so let's quickly jump into the research questions. The research questions are in three layers. The first one is about selection. We are asking questions like do female and male historical figures get included in Wikipedia based on the same criteria. Secondly, about content building. We are interested in whether there is an association between the gender and the type of information available. Thirdly, it's about the positioning in the network. We're interested in whether male and female historical characters have different structural properties in the Wikipedia link network. For each level, we want to have a look at one form of gender inequality, inclusion, stereotyping, and marginalization. The first data source was the character list in Beth's H.E., compiled by Hewner and her colleagues. It included over 500 characters. Roughly 20% of them are unnamed, like the aged priest you can see in the last item in the screenshot, and only a small fraction of them are female. Another data source is naturally the Wikipedia, and we managed to map 82% of the named characters in Wikipedia. We used the query services as listed below and got the QID, the label, the language additions, the hyperlinks on the Wikipedia pages, and a bunch of properties here. For the first research question about asymmetries in the selection process, we had to look at the notability since it was used to decide whether a given topic and in this case, a given character warrants its own article. We adopted two proxy measures of notability. One of them is the count of language additions. We used the zero-inflated negative binomial regression to investigate the gender's role on notability. The zero-inflation part tells us female characters are less likely to have any Wikipedia pages because they are more likely to be excess zeros. The negative binomial part of the model tells us the female group are more likely to be more notable in Wikipedia because they are more likely to have more language additions compared to their male counterparts. For the second question about asymmetries in content building, we did a square test for a list of properties. The major takeaway would be the male characters had better coverage in occupation, religion, or date of death, while the females had better coverage in familial relations like spouse, child, father, mother. For the third question about asymmetries in structural placement, we can only present part of our outcomes here because of limited time. We constructed a hyperlink network with each node representing a character, and then we wrecked the characters by three measures of centrality. Communicability, page rank, and betweenness attracts the fraction of female characters among the top end, which means from the top one to the full group. And then we see how this fraction of female converges or how quickly it can converge to the expected ratio, which is the actual female fraction within the whole group. Then we compare the results from this empirical graph with the results from the three constructed baseline. We randomly rewired every age. For degree sequence, we try to preserve the degree distribution of the empirical graph. In the case of small world, we constructed the network while keeping the gender ratio and also some other key factors like numbers of nodes, age, and average degree and so on. Here are the results. We can see that for the case of random and small world, it converges to the expected ratio quite rapidly for all three centrality measures we can see here. While in the case of empirical graph and the degree sequence preserving graph, it did not converge till the very end. We can also see that women consistently exhibit lower centrality in the top 50 to 100 nodes of the empirical graph than in the degree sequence preserving graph. This suggests that there are some factors beyond network structure heterogeneities that may contribute to their limited centrality in the link network. In the case of communicability, we can see a short surge in the fraction of women if we look at the top 50 nodes. This much indicates that there are some specific women that have high influence despite the overall underrepresentation of women. Some nodes may have high communicability although they are connected to only a limited number of nodes with very central roles. These nodes have limited number of direct connections to the broader network but they can still spread information and influence others through their powerful channels around them. In conclusion, women are less likely to have any Wikipedia pages but more likely to have more language additions. We know women have less information in biographical information, less information in occupation and religion while better coverage in relational information like spouse or children. For positioning, we noticed that some women have high influence while they are limited and powerful direct neighbors. We argue it is still a form of marginalization and miniature of structural barrier limiting their direct access to the broader network. Thanks for listening.