Textual risk mining for maritime situational awareness
Bibliographic record
Abstract
In this paper, we propose an auxiliary Machine Learning (ML) and Natural Language Processing (NLP) integrated system for maritime situational awareness (MSA) operations. We bring into account a new and influential asset - human intuition and perception - to the existing semi-automated decision support systems that mostly rely on numerical data collected by electronic sensors or cameras located either directly on the vessels or in the maritime command-and-control centers. For our project, we gathered weekly textual reports spanning twelve months from the United States Worldwide Threats to Shipping Reports repository that belongs to the National Geospatial-Intelligence Agency (NGA), We considered the maritime incident reports written by human operators as a valuable and accessible unstructured textual input source in which a span of text1is called “risk” if it expresses one of the following kinds of vessel incidents: fired, robbed, boarded, hijacked, attacked, chased, approached, kidnapped, boarding attempted, suspiciously approached or clashed with. Our approach benefits from probability distributions of some useful features annotated based on a list of lexicons that contain expressions denoting vessel types, risks types, risk associates, maritime geographical locations, dates and times. These distributions are captured and used to anchor the span of “risks” as they are described in the textual reports. After some preprocessing steps that include tokenization, named entity extraction and part-of-speech tagging, the textual risk mining system applies a variety of sequence classification algorithms, e.g., Conditional Random Fields, Conditional Markov Models and Hidden Markov Models in order to compare the risk classification performance. Empirical results show that our NLP/ML-based system can extract variable-length risk spans from the textual reports with about 90% correctness.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.009 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.006 | 0.003 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.003 |
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 source (direct Gemma or distilled Codex), 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".