Average call rates for US inmates are trending downward, and inmates are making more calls than ever before.
Inmates are expected to spend at least 12 billion minutes on the phone this year, based on a sample of internal numbers provided by major call providers, and most of these calls will be digitally recorded. These calls, for the most part, have no effect on anyone other than the participants, between inmates and loved ones they miss, or help them feed their pets or pay their electricity bills. Capture everyday conversations with friends you may have.
Occasionally, however, some of these inmate conversations can pose security threats to inmates, correctional staff, and the general public, and correct identification and deterrence of these threats within correctional facilities is critical. There is an urgent need. However, with new technologies such as machine learning, the power to analyze and interpret vast amounts of inmate communication data is now at our fingertips. And these digital tools may even identify threats and provide insight into inmate behavior and needs.
Flagging security threats
The use of machine learning technology has become increasingly prevalent in recent years across a wide range of industries, and criminal justice is no exception. By combining machine-learning processes with human review, correctional facilities review all inmate communications, not just a fraction, and use data-driven intelligence to enhance organizational security and corporate operations. can make decisions.
Inmates can engage in a variety of illegal activities within correctional facilities, making it increasingly important for staff to have tools to mitigate threats. One of her most commonly reported activities is extortion, in which inmates pressure their families to provide funds to other inmates, or directly threaten outsiders. Additionally, inmates may become involved in conspiracies by working with outsiders to hide evidence or falsify testimony. Inmates can also threaten facility security by smuggling contraband or planning violent escapes against correctional staff.
Currently, flagging these security threats in the piles of call recordings is a major challenge for correctional organizations. If a human were to go through all the corrective call records across the United States at normal speed, it would only take a team of 100,000 people working full time to catch up.
To alleviate that logistical hurdle, the traditional strategy employed by correctional facilities is to target a specific percentage of calls to confirm and inmates with violent or drug-related charges or gang affiliation. By focusing on calls from , you’re more likely to find actionable information. , or previous beliefs.
As a strategy, most people would agree that searching for needles in haystacks while ignoring most of them is not the best option, but correctional organizations that can’t do more than listen to the phone. For a long time, this is the only viable approach.
But machine learning can certainly improve moribund analog processes. This technology is not a distant future possibility, but a reality that is already changing the way correctional facilities review inmate communications.
How machine learning can help
Machine learning helps inmates review their communications by automatically transcribing and translating audio, scanning for secret words and phrases, creating word clouds and performing natural language processing, analyzing patterns, and providing spot audio capabilities. It can also transcribe an inmate’s phone calls to text with high accuracy, scan for keywords and names, create a visual representation of conversational topics, and analyze communication patterns to identify suspicious activity.
Additionally, by associating each word with a timecode, Call Monitor can skip to specific parts of the call recording to hear the words in context. These capabilities improve the ability of algorithms to understand and interpret communications, flagging suspicious activity and security threats for human review.
A sample process using these techniques automatically transcribes all phone calls, detects and translates foreign languages, recognizes passwords, suspicious phrases, and three-way communication indicators (i.e., if the recipient of the call When bridging, you can start by searching for recordings (commonly practiced). (prohibited in correctional facilities), and questionable metadata (calling and deposit patterns closely matching past viable calls). These initial automated processes can flag calls that are likely to be of interest to investigators.
For next-level review, the software takes flagged calls in the first step, extracts interesting words and phrases, and creates a word cloud, much like a human summarizing the call given enough time. and process call transcripts into readable summaries. Human reviewers can review the word her cloud or read transcript summaries very quickly to determine if there are any calls that need additional attention. If the percentage of calls is low, reviewers can click a keyword or phrase to hear the section of the call where that content was found. This allows reviewers to hear the selected text spoken in context. For example, a phrase like “I’m going to kill him” may imply very different things when said in a joking or friendly tone than when said outright. I have. As a final step, calls that have not yet been rejected by the previous review process can be reviewed in their entirety by either reading the full call transcript or listening to the full call recording.
With the help of machine learning, correctional facilities set themselves up to improve the inmate experience while driving positive outcomes for inmates and staff, facilitating reintegration within communities on the other side. can do. What may have once been the world of science fiction now has the potential to bring tangible change to countless individuals and communities within the criminal justice system. The momentum will only grow as technology becomes more powerful and precise.
About the author
Christopher Ditto is Vice President of Research and Development for ViaPath Technologies, a provider of inmate communications technology in the United States. Over the past 10 years, engineer, software he is an architect and project manager working on building inmate communications and tablets. We have implemented technology in correctional facilities.