Detecting Illicit Fishing Activity by Combining Open Source Data

• Activity Number: 619 - Topics in Defense and National Security
• Type: Contributed
• Date/Time: Thursday, August 1, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Karl Pazdernik* and Shari Matzner and Lauren Charles and Theodore Nowak
• Companies: Pacific Northwest National Lab and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory
• Keywords: fishing; hidden Markov model; maritime; sentiment analysis; text analysis; trending

Abstract

Fishing can be undesirable in at least three different ways. First, overfishing threatens to drive many fish species to extinction. Second, fishing gear can damage coral reefs, which support the health of underwater ecosystems. Third, fishing in contested waters can instigate hostile action between the nations in question. These concerns necessitate the need for predictive and real-time detection of illegal, unreported and unregulated (IUU) fishing activity, often identified with coordinated behavior in the water. We use trending analysis and anomaly detection in open source text data to predict where activities of interest are likely to occur. Given the narrowed potential search area of interest, we then apply a hidden Markov model to maritime Automatic Identification System (AIS) data to estimate the probability of IUU coordinated activity or vessel encounters.