The Cobwebs Company of Israel has developed a web intelligence (WebInt) collection platform for analysts in government security agencies. Udi Levi, Omri Timianker, and Shai Atias are the Company founders. Some of them had gained operational experience serving in the Israeli intelligence community.
"The advantage of the platform we developed is in the accessibility offered to the Analyst," explains Timianker. "In Israel, we are used to a situation where analysts normally possess a technical background, but that is not the case in other countries. In many cases, analysts are not technically trained, and system operation is an important point. We have seen clients who acquired WebInt systems for millions of dollars and eventually used Google to search for what they were after. Why? Because the user interface was inaccessible."
At Cobwebs, they stress that the challenge with today's WebInt collection systems involves the analysis of the data being collected. Government clients acquire intelligence collection tools that produce massive amounts of data, but the Achilles' heel is in the process of analyzing the data and drawing conclusions within a tactical timeframe. "We specialize in tactical intelligence," adds Timianker. "Our platform is intended for government security organizations. For the time being, we have not entered the field of business intelligence for the business sector – we may decide to do that in the future. We provide automation of data filtration and extraction for the analyst who may be someone with no technical code writing capabilities. The idea is to provide him with a simple search engine, like Google, for tactical Open Source Intelligence (OSInt)."
Only Open Source Intelligence
Levi stresses that the intelligence collection performed by their platform is based entirely on Open Source Intelligence (OSInt), unlike other companies in the field whose systems are based on the analysis of data in government databases or the databases of communication vendors. "Pursuant to our sales, we succeeded in understanding what the clients look for. We work strictly on collection (of intelligence) from open sources," says Levi.
"Our platform consists of three sub-systems operating together – intelligence collection, avatar management and a social engineering module that enables us to approach the target intelligently. This is a complete solution, and clients may purchase only a certain element, according to their budget.
"The system also features artificial intelligence capabilities for data fusion. Simplified, we use three methods from the discipline: classification, regression for forecasting and clustering. The classification activity enables us to identify elements within an image, like a handgun or a suitcase. This is not a new development – it has been available for years, but in the WebInt world it is an innovation. Through the use of machine learning algorithms, we no longer have to write hundreds or thousands of rules. Instead, the computer learns how a handgun or a suitcase looks, and it will identify them within the image on its own.
"The use of regression enables us to perform forecasting. One example is to characterize the behavior patterns of numerous profiles that monitor news associated with ISIS. As a result of the learning process, the computer will know how to automatically identify whether a new profile, with which it is not familiar, meets the definitions or not. Namely – it will be able to predict whether or not a specific profile belongs to a specific group of people who monitor ISIS.
"A third method is to classify the results according to predetermined parameters. This method is on the border between automation and artificial intelligence – the ability to quickly classify massive amounts of data into groups. Some of the parameters used for the classification process are derived from the other methods used to characterize profiles. In this way, the analyst, at the push of a button, can display groups and sub-groups of people around specific parameters. If he/she had to do this manually, it would have taken him/her months, and it is not at all certain that he/she would have succeeded. We are also involved in studies with Reichman – the Interdisciplinary Center in Herzliya – around the classification of terrorist profiles in the various media.
"Some of the other tools we introduced include text analysis. Other companies offer this capability, but once again, in the WebInt field, it is an innovation - including all of these capabilities in concentrated form in a single intelligence collection system. Text analysis enables the user to extract locations, names of individuals and other elements based on keywords in the text being processed. If the system identifies the word 'Paris,' for example, it will indicate a location on a map. If it identifies the name 'Donald Trump,' it will tag this expression as a name and a personality – all fully automatically. This operation is almost language-independent. If a language is missing, the system may be taught any desired language or dialect."
The ability to analyze images and texts also enables video analysis. As video is, in fact, a sequence of still images (25 or 30, depending on the format) and the audio element may be converted into text using an existing speech-text engine, the system by Cobwebs offers the option of analyzing video files as well.
"In WebInt systems, video analysis is an innovation. Our system enables the user to analyze video files, including the imagery and the audio. At the conclusion of the analysis process, the analyst will be provided with a report indicating all of the relevant events/incidents contained in the video sequence analyzed. This will significantly reduce the time required to complete the analysis process, as instead of having to watch a video sequence of one hour from surveillance cameras, he/she can complete the same task within minutes, as the entire analysis is performed automatically.
"This analysis enables the analyst to compare the images and keywords found with his/her own existing databases. The analysis also includes biometric identification of faces. The comparison against existing databases can generate alerts if someone interesting had been identified in the video sequence."
On the Agenda: The Asian Market
Cobwebs is, admittedly, an Israeli company with good connections in the local defense establishment, but the potential market for its products is mainly found overseas. The Asian market is one of the largest. The terrorist attacks in Asian countries over the last few years positioned the local security services and law enforcement agencies in an impossible situation, and they are looking for tactical capabilities, including WebInt capabilities. "We have recently opened an office in Singapore that serves the entire Asian region," says Timianker. "Next year, we intend to open an office in the USA as well. These offices were intended to deal with particularly large markets. In all of the other regions, worldwide, we operate through local partners.
"Our intelligence provides a picture of individuals, groups and areas of interest. At the bottom line, an intelligence service seeks the person at the end, the one responsible for staging or planning the attack. The objective of this intelligence effort is to catch them. In the event that a terrorist attack has taken place, you may not have succeeded in preventing it, but your analyst will still have information regarding individuals associated with it or possibly regarding the planners – all within a tactical timeframe – so you may be able to prevent follow-up attacks. That is the objective. Another objective is advance prevention. In the case of law enforcement agencies, they want to reach the perpetrators of a crime or prevent future criminal activities."
Are defense/security clients interested in adopting the new WebInt capabilities? That is a challenging subject. According to the experience gained by the people at Cobwebs, the analysts of many defense/security organizations worldwide are used to working manually and to using civilian-public systems like Google, Facebook or Twitter. "Defense/security clients, surprisingly, are only now beginning to adopt investigative automation and artificial intelligence," explains Levi.
"Perhaps in Israel the situation is better, but in other countries, including western countries, the civilian sector has been the one making the more rapid progress in the adoption of such technologies. In the defense/security sector, the clients have been more conservative, for various reasons. It could be the structured, orderly corporate culture, regulation, legislation or simply the habits of the analysts. With some analysts, you may show them a simple function like automatically finding opinion leaders around specific keywords or finding details around a telephone number – and even that will be a step forward compared to what they currently have. The lion's share of the WebInt effort in the defense/security sector is currently carried out manually. It will be a while yet before automation and artificial intelligence tools have become a part of the operational routine, and that is where we are aiming."