Needle in the Noise: Detecting Public Safety Events Over Twitter
Bibliographic record
Abstract
The recent advent of using social media and search engine queries to detect and classifying events is an emerging area in data science. This study uses current Natural Language Processing (NLP) techniques for deep learning combined with classical techniques (heuristic) to detect incidents. Individual tweets will be processed into features and correlated to detect public safety related events. These features will be derived using both classic and modern techniques. Classic features include the use of sentiment/emotion detection, word encoding, named entity recognition, and other classic semantic tools. Hybrid methods will use feature hashing to reduce features for word encoding and named entities. Modern features will use word vectors (FastText). Improved activation functions (Leaky RELU), dropout, and normalization will be used to improve performance of neural networks. Classic and Modern feature and various techinques will be assessed in terms of improving performance. Various models will be used and compared to include traditional machine learning models (Support Vector Machine, Naive Bayes) as well as deep neural networks with the goal to select models that improve performance. Classical techniques will leverage the following tools: Linguistic Inquiry and Word Count (LIWC), Valence Aware Dictionary and sEntiment Reasoner (VADER), and the National Research Council Canada's (NRC) Sentiment and Emotion Lexicons, and DBPedia Spotlight Named Entity Recognition. Detected incidents will further be classified into specific types (fire, shooting, car accident). The goal is to create a combined expert system capable of detecting incidents that impact public safety more efficiently and effectively than previous techniques and will perform better than state of the art text classification tools.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.004 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".