Context-driven result extraction system and method

1 . A system for context-driven search and result extraction in a network of content nodes, the system comprising:

an interface module configured to:

obtain a user-defined search query comprising search parameters for extracting search results from at least one target content node;

present structured search results to the user; and

obtain user feedback based on the structured search results;

a query processing module configured to process the user-defined search query and extract relevant search results using at least one trained machine learning algorithm, the query processing module comprising:

a crawler module configured to:

process the user-defined search query to extract the search parameters;

generate a query embedding based on the extracted search parameters, the query embedding comprising a vector representation of the extracted search parameters;

locate the at least one target content node in the network of content nodes;

obtain content embeddings corresponding to resources located at each target content node from a vector database, the content embeddings comprising vector representations of the resources located at the target content nodes;

determine semantic matches to the query embedding within the obtained content embeddings, wherein the semantic matches correspond to resources within the target content nodes matching the query, and the semantic matches are determined by searching the vector database for content embeddings having a similarity with the query embedding satisfying a similarity threshold; and

navigate to the matched resources identified in each target content node;

a content extraction module configured to parse each of the identified matched resources into relevant content and non-relevant content based on the corresponding content embeddings, and

a result extraction module communicatively coupled to the interface module configured to:

determine a set of fields to structure the relevant content from the matched resources; and

generate a structured search result for each of the matched resources by selectively populating the set of fields based on the content embeddings of the relevant content, wherein the query processing module is configured to receive and process the user feedback to adapt the trained machine learning algorithm.

2 . The system of claim 1 , wherein the content extraction module is configured to:

identify, within the matched resources, nested links corresponding to additional resources of the at least one target content node; and

communicate the nested links to the crawler module for determining additional semantic matches.

3 . The system of claim 1 , wherein the result extraction module is configured to:

define weights for each field of the set of fields; and

score each of the structured search results based on the weights of the populated fields.

4 . The system of claim 1 , wherein the result extraction module is configured to:

generate a result embedding for each of the structured search results;

determine a similarity of each result embedding with the other result embeddings; and

remove duplicate structured search results for result embeddings having a similarity with each other greater than a predetermined threshold.

5 . The system of claim 1 , wherein the crawler module uses a first trained machine learning algorithm, wherein the content extraction module uses a second trained machine learning algorithm, and wherein the result extraction module uses a third trained machine learning algorithm.

6 . The system of claim 5 , wherein the interface module is configured to dynamically select a preferred trained machine learning algorithm from a plurality of trained machine learning algorithms to use for one or more of the first trained machine learning algorithm, the second trained machine learning algorithm, and the third trained machine learning algorithm based on the search parameters.

7 . The system of claim 6 , wherein the interface module comprises an agentic artificial intelligence agent configured to select the preferred trained machine learning algorithm to use for one or more of the first trained machine learning algorithm, the second trained machine learning algorithm, and the third trained machine learning algorithm based on the search parameters.

8 . The system of claim 1 , wherein the interface module is configured to communicate with a search engine to select the at least one target content node based on the search parameters.

9 . The system of claim 1 , wherein the crawler module is configured to utilize a retrieval-augmented generation process to:

generate the query embedding;

search the vector database for the content embeddings based on a similarity with the query embedding; and

determine the semantic matches corresponding to the matched resources based on the content embeddings having a similarity with the query embedding greater than the similarity threshold.

10 . The system of claim 1 , wherein adapting the trained machine learning algorithm comprises adjusting one or more weights of the trained machine learning algorithm based on the user feedback.

11 . A method for context-driven search and result extraction in a network of content nodes, the method comprising:

obtaining, by an interface module, a user-defined search query comprising search parameters for extracting search results from at least one target content node;

processing, by a query processing module, the user-defined search query to extract relevant search results using a trained machine learning algorithm, wherein the processing comprises:

extracting, by a crawler module, the user-defined search query to extract the search parameters;

generating, by the crawler module, a query embedding based on the extracted search parameters, the query embedding comprising a vector representation of the extracted search parameters;

locating, by the crawler module, the at least one target content node in the network of content nodes;

obtaining, by the crawler module from a vector database, content embeddings corresponding to resources of each target content node, the content embeddings comprising vector representations of the resources located at the target content nodes;

determining semantic matches to the query embedding within the obtained content embeddings, wherein the semantic matches correspond to resources within the target content nodes matching the query, and the semantic matches are determined by searching the vector database for content embeddings having a similarity with the query embedding satisfying a similarity threshold;

navigating, by the crawler module, to the matched resources of each target content node corresponding to the semantic matches of the query embedding;

parsing, by a content extraction module, each of the identified matched resources into relevant content and non-relevant content based on the corresponding content embeddings;

determining, by a result extraction module, a set of fields to structure the relevant content from the matched resources; and

generating, by the result extraction module, a structured search result for each of the matched resources by selectively populating the set of fields based on the relevant content;

presenting, by the interface module, the structured search results to a user;

obtaining, by the interface module, user feedback based on the structured search results; and

adapting, by the processing module, the trained machine learning algorithm based on the user feedback.

12 . The method of claim 11 , further comprising:

identifying, by content extraction module, nested links within the matched resources corresponding to additional resources of the at least one target content node; and

communicating, by content extraction module, the nested links to the crawler module for identifying additional semantic matches.

13 . The method of claim 11 , further comprising:

defining, by the result extraction module, weights for each field of the set of fields; and

scoring, by the result extraction module, each of the structured search results based on the weights of the populated set of fields.

14 . The method of claim 11 , further comprising:

generating, by the result extraction module, a result embedding for each of the structured search results;

determining, by the result extraction module, a similarity of each result embedding with the other result embeddings; and

removing, by the result extraction module, duplicate structured search results for result embeddings having a similarity with each other greater than a predetermined threshold.

15 . The method of claim 11 , wherein the crawler module uses a first trained machine learning algorithm, wherein the content extraction module uses a second trained machine learning algorithm, and wherein the result extraction module uses a third trained machine learning algorithm.

16 . The method of claim 15 , further comprising:

dynamically selecting, by the interface module, a preferred trained machine learning algorithm from a plurality of trained machine learning algorithms to use for one or more of the first trained machine learning algorithm, the second trained machine learning algorithm, and the third trained machine learning algorithm based on the search parameters.

17 . The method of claim 16 , wherein the interface module comprises an agentic artificial intelligence agent configured to select the preferred trained machine learning algorithm to use for one or more of the first trained machine learning algorithm, the second trained machine learning algorithm, and the third trained machine learning algorithm based on the search parameters.

18 . The method of claim 11 , further comprising:

communicating, by the interface module, with a search engine to select the at least one target content node based on the search parameters.

19 . The method of claim 11 , wherein the crawler module is configured to utilize a retrieval-augmented generation process comprising:

generating the query embedding;

searching the vector database for the content embeddings based on a similarity with the query embedding; and

determining the matched resources based on the content embeddings having a similarity with the query embedding greater than the similarity threshold.

20 . A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause a computing system to:

obtain a user-defined search query comprising search parameters for extracting search results from at least one target content node of a network of content nodes;

process the user-defined search query and extract relevant search results using a trained machine learning algorithm, wherein the instructions to process the user-defined search query cause the computing system to:

process the user-defined search query to extract the search parameters;

generate a query embedding based on the extracted search parameters the query embedding comprising a vector representation of the extracted search parameters;

locate the at least one target content node in the network of content nodes;

obtain content embeddings corresponding to resources of each target content node from a vector database, the content embeddings comprising vector representations of the resources located at the target content nodes;

determine semantic matches to the query embedding within the obtained content embeddings, wherein the semantic matches correspond to resources within the target content nodes matching the query, and the semantic matches are determined by searching the vector database for content embeddings having a similarity with the query embedding satisfying a similarity threshold;

navigate to the matched resources of the target content nodes;

parse each of the identified matched resources into relevant content and non-relevant content based on the corresponding content embeddings;

determine a set of fields to structure the relevant content from the matched resources; and

generate a structured search result for each of the matched resources by selectively populating the set of fields based on the content embeddings of the relevant content;

present the structured search results to the user via a user interface;

obtain user feedback based on the structured search results; and

adapt the trained machine learning algorithm based on the user feedback.