Compare text to a standard set of terms - Deep Learning, NLP
- Status: Closed
- Prize: $200
- Entries Received: 3
- Winner: nitinbhardwaj01
We are a startup that specializes in AI solutions for businesses. Our fast growing business works in the legal-, real estate- and services industry. For the area ML, DL Vision and NLP ML and DL we are searching for talents who can help us solve challenges like the following:
Compare any document to a standard. Let's say that you have the need to compare dozens of supplier submissions to a standard that you defined in your company in order to see whether the submissions are compliant or not. This could be a security policy, a code of conduct or any other standard. So you have a word document that defined topic by topic what your standard is. The suppliers are responding to your requirements in free text format.
Now, an AI system shall compare the free text submissions against the defined standard and report IF and TO WHICH EXTENT the submission(s) are compliant with the standard.
- Standard: All communication between the supplier data center and the customer data center have to be encrypted.
- Customer 1 submission: We use a Layer 1 encryption between on the WAN connection between data centers.
- Customer 2 submission: Our WAN connection partner uses an optional encryption service on their WAN connection lines.
- Customer 3 submission: Supplier uses xyz company product to encrypt all data going out of the supplier DC LAN.
The AI should report:
- Customer 1: Compliant
- Customer 2: Non Compliant
- Customer 3 Compliant.
Obviously the AI would have to be trained for this.
What would be your solution to address this challenge ?
- Software technologies you would use,
- Approaches for training your would use,
- Data science approaches you would use,
- Potentially some specific way to prepare the definition of the "standard" you would recommend,
You can assume that the submissions are all nice and clean files with perfect OCR. And you can assume that there are hundreds of training examples available to train the AI.
Examples of security policies can be found on the web, e.g.
[login to view URL]
Subsequent frequent hiring of talents submitting a real good contest is very likely.