rc11-group4

[rc11-group4]

Track people's movement and face to calculate their relative distance.

[rc11-group4]

Cluster analysis

Key Narrative Themes in Cluster 6

Revenge and Justice: Orestes’ matricide and the cycle of vengeance culminating in divine judgment.

Fate and Sacrifice: Prometheus’ suffering, Philoktetes’ reluctant return, and Alcestis’ salvation.

Divine Intervention: Frequent appearances of gods influencing human actions and moral outcomes.

Rhetoric and Deception: The power of speech in persuasion, manipulation, and Socratic argument.

Civic Crisis and Collective Appeals: Thebes’ plague and the search for leadership and healing.

Dramatic texts were aggregated into narrative clusters via spatial clustering analysis.

[rc11-group4]

Dramatic texts embedded within the narrative field.

Narrative texts were matched to points within the vector field based on their five-dimensional structural vectors, using Euclidean distance as the similarity metric.

These spatial coordinates reflect a certain degree of narrative resonance between drama and news. They serve as echoes of stories, myths, and legends—subtle imprints of narrative forms projected onto the contemporary urban fabric.

[rc11-group4]

A narrative field was constructed using Gaussian kernel expansion.

The structural vectors of news nodes were diffused using a 250-meter radius.

In the left figure, the color patches represent the distribution of vectors after dimensionality reduction to three dimensions, with colors reflecting the projected components.

In the right figure, the vectors were reduced to two dimensions and visualized as arrows, where color indicates vector intensity, calculated as the sum of five selected dimensions.

[rc11-group4]

Structural vectors were embedded into urban space

The connectivity of the SVO network was projected onto the map, enabling a spatial interpretation of narrative relationships.

The structural vector of each news event was visualized using PCA.

[rc11-group4]

Structural metrics were computed for the nodes in the network

The computed structural metrics include the out-degree and in-degree of subjects, the frequency of objects, and the out-degree and in-degree of verbs.

These indicators help characterize how events are represented and distributed within the network.

[rc11-group4]

The network of news items

The news SVO node network was reconstructed using Pyvis, resulting in a significantly more complex structure.

[rc11-group4]

The news dataset is processed through ChatGPT and the google Maps API

The news dataset has been expanded to 6,433 entries. GPT was used to extract subject-verb-object (SVO) structures and potential location information.

The geographic coordinates (latitude and longitude) for each news item were retrieved using the Google Maps API.

[rc11-group4]

1.Theatrical City

This project adopts "dramatization" as a counterforce to intervene in the commodification-driven trends of urban space, aiming to revitalize the social vitality and public diversity of the city through the reconstruction of democratic semantics.

[rc11-group4]

2. Greek drama dataset

Source: 18 ancient Greek dramas. Data volume: approximately 2.1 million words.

[rc11-group4]

2.Greek drama dataset

Use the GPT-4o API to split sentences into SVO (Subject-Verb-Object) structures.

[rc11-group4]

2. Athens News dataset

3,567 modern news articles (Google news), with SVO triples extracted using Stanford NLP techniques.

[rc11-group4]

2.News and Greek drama dataset

Extract features by training data using a Self-Organizing Map (SOM).

[rc11-group4]

3.Network visualization and community detection

Use the Louvain algorithm to segment the news data into 10 networks.

[rc11-group4]

3.Network visualization and community detection

There are 18 networks in total for the scripts.

[rc11-group4]

4.Generate 3D coordinates based on verb vectors

Use the BERT model to vectorize the V (verb) in SVO, Apply PCA to reduce the dimensionality to 3D coordinates (x, y, z).

[rc11-group4]

5.Match the news network with the script network

Find the most similar news network for each script by comparing using Euclidean distance.