AAI_2025_Capstone_Chronicles_Combined

Cinema Analytics and Prediction System

9

Figure 7: Cinema Analytics and Prediction System Architecture

Application Research and Machine Learning Method

Application Research

The project’s analytical strategy is divided into two main components: Natural Language

Processing (NLP) tasks and predictive modeling tasks. The NLP component involves text-based

analyses such as genre classification and content-based movie recommendation, focusing on

extracting and interpreting semantic features from movie descriptions. The predictive modeling

component focuses on revenue prediction and success classification. Revenue prediction is

framed as a regression problem aimed at estimating a continuous target variable, while success

classification is approached as a classification problem to determine whether a movie is likely to

be a hit or a flop. Together, these components provide a comprehensive and holistic framework

for analyzing the movie industry from both linguistic and numerical perspectives.

The genre classification task involves predicting one or more genres (multi-label)

associated with a movie using a combination of its textual descriptions and numeric metadata.

The recommendation task is designed to suggest similar or relevant movies to users based on

the semantic content of the movie.

There are several well-established methods and technologies used to address similar

problems in real-world and academic contexts. Commercial systems such as Netflix, IMDb, and

176

Internal