Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Identifying Risk Patterns in Brazilian Police Reports Preceding Femicides: A Long Short Term Memory (LSTM) Based Analysis

Vinicius Lima, Jaque Almeida de Oliveira

TL;DR

This paper addresses predicting risk and potential escalation to femicide by analyzing Brazilian police reports with Long Short-Term Memory networks. It implements two LSTM-based approaches: (i) a text-classification model that labels reports as Higher Risk or Lower Risk with 66% accuracy, and (ii) a sequence-prediction model that forecasts the next action in a violence progression using manually extracted features. The study uses 162 reports from 2017–2021 tied to 142 femicide cases, with 39 high-risk and 79 low-risk instances preceding femicide, highlighting the potential of NLP to support law enforcement in risk assessment and prevention. While promising, the work notes limitations due to small data, potential biases, and reliance on manual feature extraction, calling for larger datasets and attention mechanisms to improve robustness and applicability.

Abstract

Femicide refers to the killing of a female victim, often perpetrated by an intimate partner or family member, and is also associated with gender-based violence. Studies have shown that there is a pattern of escalating violence leading up to these killings, highlighting the potential for prevention if the level of danger to the victim can be assessed. Machine learning offers a promising approach to address this challenge by predicting risk levels based on textual descriptions of the violence. In this study, we employed the Long Short Term Memory (LSTM) technique to identify patterns of behavior in Brazilian police reports preceding femicides. Our first objective was to classify the content of these reports as indicating either a lower or higher risk of the victim being murdered, achieving an accuracy of 66%. In the second approach, we developed a model to predict the next action a victim might experience within a sequence of patterned events. Both approaches contribute to the understanding and assessment of the risks associated with domestic violence, providing authorities with valuable insights to protect women and prevent situations from escalating.

Identifying Risk Patterns in Brazilian Police Reports Preceding Femicides: A Long Short Term Memory (LSTM) Based Analysis

TL;DR

This paper addresses predicting risk and potential escalation to femicide by analyzing Brazilian police reports with Long Short-Term Memory networks. It implements two LSTM-based approaches: (i) a text-classification model that labels reports as Higher Risk or Lower Risk with 66% accuracy, and (ii) a sequence-prediction model that forecasts the next action in a violence progression using manually extracted features. The study uses 162 reports from 2017–2021 tied to 142 femicide cases, with 39 high-risk and 79 low-risk instances preceding femicide, highlighting the potential of NLP to support law enforcement in risk assessment and prevention. While promising, the work notes limitations due to small data, potential biases, and reliance on manual feature extraction, calling for larger datasets and attention mechanisms to improve robustness and applicability.

Abstract

Femicide refers to the killing of a female victim, often perpetrated by an intimate partner or family member, and is also associated with gender-based violence. Studies have shown that there is a pattern of escalating violence leading up to these killings, highlighting the potential for prevention if the level of danger to the victim can be assessed. Machine learning offers a promising approach to address this challenge by predicting risk levels based on textual descriptions of the violence. In this study, we employed the Long Short Term Memory (LSTM) technique to identify patterns of behavior in Brazilian police reports preceding femicides. Our first objective was to classify the content of these reports as indicating either a lower or higher risk of the victim being murdered, achieving an accuracy of 66%. In the second approach, we developed a model to predict the next action a victim might experience within a sequence of patterned events. Both approaches contribute to the understanding and assessment of the risks associated with domestic violence, providing authorities with valuable insights to protect women and prevent situations from escalating.
Paper Structure (10 sections, 3 figures, 3 tables)

This paper contains 10 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Objectives
  4. Methodology
  5. Dataset
  6. Preprocessing and Modeling
  7. Results and Discussion
  8. LSTM for identification of higher or lower risk.
  9. LSTM for sequence behavior prediction.
  10. Conclusion

Figures (3)

  • Figure 1: Violent-o-meter scale created by IPN ipn_portal.
  • Figure 2: Femicide cases in the Federal District of Brazil.
  • Figure 3: Word cloud (in Portuguese) resulted from the 162 reports.