In this episode, I will answer questions about how artificial intelligence powers Cellebrite Pathfinder to provide quick highlights and relevant insights during investigations.

How does the latest artificial intelligence research on natural language understanding affect digital intelligence?

Most of the data we collect from digital devices is unstructured in nature. In a typical case, we may be talking about millions of images, videos, text messages, and documents. Manual review of this much data is impossible, which is why solutions (powered by artificial intelligence), have become so necessary in aiding modern investigations.

Previously, obtaining insights from text was limited to keyword or watchlist extraction, and (if you were lucky), pattern, and entity extraction. However, recent developments in transformer-based, neural-network architectures, allow computers to capture a lot more of the context and meaning in written text, which can help investigators zero in on valuable data.

To learn more about the value of AI and see real-life cases in action, download our Whitepaper, “How AI and Machine Learning Are Impacting Digital Investigations.” View real-life cases that explain how image categorization made a difference in their investigations.

How does Cellebrite Pathfinder utilize this technology?

Cellebrite Pathfinder uses an optimized, multi-lingual, transformer-based language model to analyze individual chat messages and complete chat conversations.

It can capture the meaning of a message and flag whether it is related to topics like drugs, money, law enforcement, intent-to-delete data, and many more topics.

It can also capture higher-level concepts for entire conversations and potentially identify child exploitation, drug deals, and online scams.

The technology goes beyond keywords and watchlists and can identify those concepts even in cases where slang or innuendo are used … and… it can do so in dozens of different languages.

Learn more about how Cellebrite Pathfinder can expedite your investigations, here.

Expert Profile:

Oren Yosifon has over 15 years of hands-on coding, software architecture, distributed systems design, text analytics, information retrieval, social-network-analysis, web-mining, and semantic data integration experience in combination with a strong IT background.

Oren enjoys creating robust, scalable technologies and algorithms enabling users to fuse and unlock the potential of structured and unstructured data. The main languages he uses are Java, Groovy, Scala, Javascript, Python, R, and C#.

Share this post