Systems and methods for determining financial security risks using self-supervised natural language extraction

A method for dynamic detection of security features based on self-supervised natural language extraction from unstructured data sets, comprising:

• retrieving a first unstructured data array indicative of a plurality of first discrete user events, wherein retrieving the first unstructured data array comprises retrieving the first unstructured data array from a database;
• converting the unstructured data array by serializing the unstructured data array based on at least one indicator to form one or more first data arrays, the one or more first data arrays each indicative of a discrete user event;
• for each data entry of the one or more first data arrays, determining a tokenization library using a neural network implemented by one or more processors;
• generating a value for each data entry of the one or more first data arrays based on the determined tokenization library, wherein the value for each data entry is also determined at least in part by accessing the database to find the respective first discrete user event for each data entry to determine whether the value represents a desired event with a positive value or a hazardous event with a negative value;
• determining one or more second data arrays corresponding to a subset of the one or more first data arrays based on selecting a plurality of highest value data entries from the one or more first data arrays using the neural network, wherein selecting the plurality of highest value data entries includes dynamically generating narratives associated with each respective first discrete user event of the one or more first data arrays and storing the narratives as the one or more second data arrays in the database;
• determining a security weight for each discrete user event based on the one or more second data arrays and the at least one indicator associated with a respective discrete user event, the at least one indicator comprising a timestamp;
• determining an associated sentiment score for each security weight;
• determining a security score based on a weighted average of the sentiment score and the security weight;
• when the security score of any discrete user event exceeds a predetermined threshold, executing one or more security actions, wherein at least one of the one or more security actions comprises generating an indication that the security score of a user account associated with the discrete user events exceeds the predetermined threshold;
• computing one or more third data arrays from the one or more second data arrays using the neural network by providing the one or more second data arrays as input to the neural network;
• iteratively calculating an error measurement between the one or more third data arrays and the one or more first data arrays;
• and
• iteratively modifying one or more weights of one or more layers of the neural network based on the calculated error measurement.

A system for dynamic detection of security features based on self-supervised natural language extraction from unstructured data sets, comprising:

• one or more processors;
• a memory in communication with the one or more processors and storing instructions that, when executed by the processor, are configured to cause the system to:
• retrieve a first unstructured data array comprising a full-text corpus indicative of one or more financial assets narratives and one or more financial liabilities narratives associated with a user, wherein retrieving the first unstructured data array comprises retrieving the first unstructured data array from a database;
• serialize the first unstructured data array based on a pickling algorithm to form one or more first data arrays each indicative of a portion of the full-text corpus, wherein each portion is indicative of a financial assets or a financial liability narrative and includes an associated timestamp, wherein the portions of the narrative are classified as financial assets or financial liabilities at least in part by accessing a user information database which includes semantic data indicative of whether a financial event in the full-text corpus is a desired positive event or a hazardous negative event;
• apply a self-tokenization aspect of a natural language processing algorithm to determine key words and nearby word correlations for each word contained in the full-text corpus using a neural network implemented by the one or more processors, wherein determining key words and nearby word correlations includes determining which words of the full-text corpus convey an outcome of the one or more financial assets narratives or one or more financial liabilities narratives, wherein the one or more financial assets narratives and the one or more financial liabilities narratives are stored in the database;
• determine, using a self-supervised machine learning aspect of the natural language processing algorithm, a condensed summary of each portion of the full-text corpus corresponding to a subset of the one or more first data arrays;
• determine a relevancy score for each condensed summary based at least in part on the associated timestamp;
• determine, using a sentiment analysis aspect of the natural language processing algorithm, a sentiment score for each condensed summary;
• determine an overall risk score for each condensed summary based on a weighted average of the relevancy score and the sentiment score;
• when at least one overall risk score exceeds a predetermined threshold, execute one or more security actions, wherein the one or more security actions comprise at least flagging a user account for external review;
• generate a reconstructed full-text corpus by providing the condensed summaries of each portion of the full-text corpus as input to the neural network;
• iteratively calculate an error measurement between the reconstructed full-text corpus and the full-text corpus;
• and
• iteratively modify one or more weights of one or more layers of the neural network based on the calculated error measurement.