Likes, shares and drug deals: WVU researchers create model that uncovers illegal drug trafficking on social media | WVU today – 71Bait

Social media can be a lot more than political tirades and snaps of delicious meals or furry friends.

West Virginia University researchers have found that social networking platforms can serve as a direct-to-consumer marketing tool for drug dealers to sell illegal drugs.

Professor Xin Li of the Lane Department of Computer Science and Electrical Engineering and Chuanbo Hu, a postdoctoral fellow, said that exposing illegal online drug trafficking has become crucial to combating drug trafficking on cyber platforms.

“Illegal drug trafficking has evolved along with technology over the past few decades,” Li said. “Social media has become a tool used not only by ordinary people but also by drug dealers.”

A popular service that drug dealers have exploited is Instagram, which Li says replaced Twitter as the primary platform for illicit drug trafficking around 2019. Compared to other social media platforms, including Tik-Tok, the algorithms associated with Instagram enable more personalized content to be delivered directly to people expressing interest in specific posts and hashtags.

Anyone who follows a dealer’s account or likes a dealer’s post causes the Instagram algorithm to populate that person’s feed with more of those drug-related posts.

To counteract this, Li, Hu and their team conducted the first systematic study of detailed detection of illegal drug trafficking on Instagram. They proposed a deep multimodal multilabel learning approach to detect IDTEs and demonstrate its effectiveness using a newly created dataset called multimodal IDTE.

This model takes text and image data as input and combines multimodal information to predict multiple illicit drug labels. Their research was presented at an international conference in late 2021.

“The multi-modal IDTE is the developed AI model to automatically detect illegal drug trafficking on Instagram,” said Hu. “Compared to previous work, this method fully considers the evidence of drug trafficking provided by images and text and realizes the fine-grained classification of drug trafficking events.”

According to Li and Hu, drug dealers on Instagram generally use hashtags to extend their reach and engage their audience, which can be attached to posts. Dealers almost always link many drug-related hashtags to increase the visibility of their posts.

Accurately detecting IDTEs from social media has become increasingly difficult due to the inconsistency of drug legislation, the vastness of social media, and the ambiguity surrounding which drugs are being posted and why.

“They encourage illicit drug trafficking in two ways: by posting a message, like an ad for a specific drug, and by replying to an existing post,” Li said. “They use slang, street names of drugs, or other methods like misspellings to avoid getting caught.”

“Some drug dealer accounts never post pictures, but only comment on some hot posts to improve the visibility of their ads,” Hu said.

Li said detecting IDTE is like finding a needle in a haystack due to the sheer size of social media data.

“Drug dealers use various tricks to avoid detection,” Li said. “The line between IDTE and regular events is not always clear. For example, someone’s grandmother might use a particular opioid as a prescription drug to treat pain.”

In contrast to existing work on detecting drug dealers or drug use from aggregated information, Li and Hu focus on detecting activities related to suspicious IDTEs. Her work also focuses on an approach that not only detects illicit drugs but also their specific types in each suspect IDTE.

Specifically, Li and Hu’s model takes text and image data associated with suspect IDTEs and puts the multimodal information together to predict multiple labels of an illicit drug.

“We refined the classification of IDTEs into nine categories, most of which include the widely used illicit drugs,” Hu said. “Taking a post with image and caption or comment as input, the proposed model can learn drug characteristics based on multimodal analysis to determine if drugs are illicit and to which category they belong.”

Researchers manually generated a large-scale multimodal IDTE dataset (MM-IDTE) for the purpose of detecting illicit drugs. The MM-IDTE dataset, which contains nearly 4,000 articles and more than 6,000 comments, represents the largest multimodal (text and image) dataset on illicit drug detection to date.

To create such a large dataset, the researchers developed an automatic data-crawling system for Instagram that uses hashtag and image information together to drive data collection.

“Manually collecting Instagram data is an impossible task,” Li said. “My team developed a data crawling system to automatically download all data (posted text and images) from Instagram. It collects raw materials to support our data mining research.”

“The automatic data crawling system is designed to collect many training samples, which are very useful to help the model automatically learn discriminative representations for classification,” Hu said. “In other words, the larger and more diverse the data, the more accurate and robust the model will be.

“This data-crawling system automatically retrieves posts from drug dealers based on the collected drug-related hashtags,” Hu said. “To improve fetching efficiency, the system automatically filters out some irrelevant posts through the AI ​​model.”

The newly created MM-IDTE dataset will be made publicly available to support research related to illicit drug trafficking activities.

Li and Hu propose a Deep Multimodal and Multilabel Learning (DMML) framework to detect illicit drug trafficking events as it can realize fine-grain classification of IDTEs considering differences in drug legalization.

“The proposed method can automatically learn distinctive features from multi-model data and recognize various drug trafficking patterns based on the proposed comment-based detection unit,” Hu said.

Li and Hu said their method can successfully identify some challenging cases that are difficult for untrained eyes, such as: B. Special icons and style changes that try to evade detection. The developed system could also facilitate law enforcement’s disruption of illicit drug trafficking.

Li and Hu believe their research can help identify different types of drug trafficking events on social media platforms.

“Our approach is not limited to Instagram,” Hu said. “It is a general tool for detecting illicit drug trafficking events by fusing multimodal data such as images and text. The reason for this design is the flexible extension to other platforms.”

“Further extensions to trade in people, natural resources and virtual products are possible,” Li said.

Minglei Yin, a graduate student in the Lane Department of Computer Science and Electrical Engineering, accompanied Li and Hu on the study.

Citation: https://dl.acm.org/doi/pdf/10.1145/3459637.3481908

-WVU-

from 04.06.22

CONTACT: Paige Nesbit
Statler College of Engineering and Natural Resources
304-293-4135; Paige.Nesbit@mail.wvu.edu

Call 1-855-WVU-NEWS for the latest West Virginia University news and information from WVUToday.

Follow @WVUToday on Twitter.

Leave a Comment