How Our Data Looks

cytoscape network as svg file — Below is a .svg image file showing some of our data being processed through cytoscape. At the center is a node labeled "Gwyn", surrounded by and connected to various item nodes. This shows that each of these items mentions Gwyn in their descriptions, or has some connection with Gwyn which makes sense, given some context. In Dark Souls, Gwyn is a central character who ultimately causes the events of the first game. After killing the dragons, Gwyn built the kingdom of Lordran and began the age of fire. When the first flame began to fade, Gwyn sacrificed himself to prolong it which put a curse on humanity. He and his actions are tied to just about everything in the game, so it makes sense he's in the center, connected to every item.

This network functions as a way to uncover information on what Natural Language Processing(NLP) programs like spaCy can accomplish. Our network focusing on gwyn allows one to see the association bewteen him and these items within Dark Souls. Using this analysis, we can interpret how AI chatbots read and understand this data, allowing for further speculation. Our project started as an idea to have chatGPT use in-game item descriptions to generate original, lore-friendly items. Seeing how AI interprets the data confirms that this is possible as it's able to make connections bewteen descriptions. Using AI can also be useful for people trying to understand some of the more hidden lore bits from Dark Souls, as the AI points out connections. Seeing Gwyn's relation to each item gives us a unique prespective that allows us to notice how interconnected each item is to the game. To an extent, this how we can understand what ai models are seeing.

Cytoscape Visualization

Next is our cytoscape visualization. Here, we have a large portion of visible connections all throughout the information present in our source document. This visualization helps piece together fully how each bit of information is connected. This includes the connections of similarly named items, such as armor pieces of the same set, and other things, such as cardinal numbers that are present in items of indexed values. By looking through this visualization, one can see more of how the information is organized, and how the ai may be reaching in to look at it. Key points of interest involve the related items, important people and places, understanding the differences between differently colored and shaped icons, and anything else that could provide an in-depth look into our dataset. Beneath the cytoscape visualization is instructions that help navigate the network.

In order to properly navigate the Cytoscape network, you should take a second to follow these instructions here. First, the visualization was meant to have a dark background. The text is white, so changing the background to black can help with reading the visualization. On the right side of the window, select the ink icon (this may take a few clicks as it may be unresponsive), and select the desired color. After this, it is now a matter of altering the zoom so that navigating around the window is possible. Simply use the scroll of your mouse to zoom in, and if you need to pan across the window, click the left and mouse button at the same time and hold for a moment. This will allow you to pan across the screen. Finally, in the top left corner, there are a few optional views that can be accessed, however it is recommended that you keep the default settings, as we personalized it for this view. It is encouraged that you explore the network and try to find associations between items. We believe this is the best way to try and understand the data and how ai views related topics!