Tauchain progress and the automation of Websluething
Recently there has been major progress with TML. Ohad has released an optimized and fully functioning meta language which can be used to define new languages. This ability is critical for Tauchain because Tauchain will require in specific KRR (knowledge representation reasoning) languages. One example of such a KRR is Attempto Controlled English, and any other controlled language can be implemented using TML. At this moment in time Ohad is working on the parser and in particular the lexical analysis/syntax analysis portion. All of this can be witnessed by observing the commits on Github.
For those who study the topic you would know the name Doug Lenat. To fully understand what Tauchain could be capable of, it helps to watch this video:
In the video it is mentioned that terrorism can be detected. This is in fact possible to implement over Tauchain and I'll discuss the idea below.
The Sherlock Holmes Bot
The required elements to have a Sherlock Holmes Bot is to have both a common sense knowledge base and reasoning capability which can connect logical dots between events. Also necessary is to have a web sensing mechanism similar to the project Websensors.
Sherlock Holmes in specific used deductive reasoning to solve crimes. If an event happens somewhere in the world then the data in most cases is uploaded to the Internet. This could be news articles, it could be Twitter posts, it could be photographs or something else. The sensing agent would simply provide the service of scanning the web detecting different events. It could for example detect news articles as they are uploaded and capture relevant topics.
In the situation where there is a news article about a pipe bomb or a suspicious explosion then the sensing agent would be able to detect this in near real time. The data then once captured could be fed into Tauchain where the reasoning capability and knowledge base could provide an investigative function. In the case of terrorist attacks the logical dots would be connected by way of deductive reasoning data such as the chain of events. Sentiment data from social media could also be captured by way of websensor agents which could again be a factor.
Of course there are some problems such as can you trust the news sources? If the news sources can be trusted to provide accurate data about real world events then Tauchain could provide a relevant result under this assumption. This would provide a use case where the knowledge base of Tauchain could be used for anti-terror/crime fighting in an automated capacity.
Examples:
Events which seem connected. Tauchain may be able to help with this too by connecting events, places, apply common sense reasoning..
Such as a certain person wearing certain clothing spotted at a certain place (from trusted data source) and also a certain person, wearing the exact same clothing, spotted at another place (from another trusted data source), in a time frame where it is logical that it could be the same person.
Tauchain would be able to tell you it's possible to be the same person due to the same clothes, and in locations near enough to each other.
The main problem is you have to trust the data sources fully. Fake data sources can make the results you get useless. If for example data sources lie about what they see then the deductive reasoning would produce a result based on lies. If the data sources tell the truth then the deductive reasoning would produce a result which is relevant to solving the crime or potentially preventing the next terrorist attack.
Conclusion
Law enforcement, anti-terrorism, all of this can be represented in Tauchain. First, there could be development of an anti-terrorism knowledge base or and anti-crime knowledge base. The Dept of Homeland Security, Interpol, or any other similar agency could populate these knowledge bases, sharing knowledge to achieve shared objectives. Tauchain would be able to reason over any knowledge base just as Cyc currently is able to do.
What Tauchain can't do (in current design) is provide sensors, or provide trust. In other words if you have trusted sensors and data you can trust then you can add it to a knowledge base to reason over it. If you cannot determine what the facts are then no amount of reasoning or logic can fix that.
All people interested in fighting crime or terrorism should be able to see the immediate potential of this. If you have one shared knowledge base on Tauchain with all relevant knowledge about terrorist or criminal groups, their members, leaders, ideology, founders, sponsors, affiliations, facilities, locations, finances, capabilities, intentions, behaviors, tactics, and full descriptions of specific terrorist events, and then you can use automated reasoning over this shared data? In my opinion there is quite a bit of potential there.
References
Deaton, C., Shepard, B., Klein, C., Mayans, C., Summers, B., Brusseau, A., ... & Lenat, D. (2005, May). The comprehensive terrorism knowledge base in cyc. In Proceedings of the 2005 International Conference on Intelligence Analysis.