AAI_2025_Capstone_Chronicles_Combined
Cinema Analytics and Prediction System
22
Figure 19: BERT-based movie cluster
Figure 20: BERT Genre classification example
The LSTM + GloVe model provided a thematically diverse yet coherent set of
recommendations. For example (Figure 21), for Transformers, the top recommendation,
Spider-Man 2 , scored 0.929 for similarity and shares high-budget action, a heroic lead, and
themes of responsibility and transformation. Left Behind (0.926) and Dragonball Evolution
(0.916) align with its apocalyptic or sci-fi elements, while Cars (0.914) matches through fast
paced action and machine-related themes. Dinner for Schmucks (0.911) appears as an outlier,
likely included due to overlapping keywords or metadata rather than strong thematic ties.
189
Internal
Made with FlippingBook - Share PDF online