Systems, devices, and methods of the present invention create a searchable index that includes informative portions of text. In an example, a computer-implemented method creates a discourse tree from a body of text. For each non-terminal node in the discourse tree, the method identifies a rhetorical relationship associated with the non-terminal node. The method labels each terminal node associated with the non-terminal node as either a nucleus or a satellite. The method further accesses a rule associated with the rhetorical relationship, and selects, based on the rule, selects the fragment associated with the nucleus. The method creates a searchable index including the selected fragments.

Methods and system for determining a similarity relationship between a plurality of digital assets and a target digital asset comprises creating a normalized semantic feature vector associated with a search query, discovering the target asset based on the normalized semantic feature vector, generating a normalized latent feature vector associated with the target asset, comparing the normalized semantic feature vector with semantic feature vectors for each of the digital assets to generate a semantic comparison value, comparing the normalized target latent feature vector with latent feature vectors for each of the digital assets to generate a latent comparison value, blending the semantic comparison vector value with the latent feature comparison vector value to create a target comparison value for each of the digital assets, and reporting the digital assets having the highest target comparison values to the user or group of users.

Methods and system for determining a similarity relationship between a plurality of digital assets and a target digital asset comprises creating a normalized semantic feature vector associated with a search query, discovering the target asset based on the normalized semantic feature vector, generating a normalized latent feature vector associated with the target asset, comparing the normalized semantic feature vector with semantic feature vectors for each of the digital assets to generate a semantic comparison value, comparing the normalized target latent feature vector with latent feature vectors for each of the digital assets to generate a latent comparison value, blending the semantic comparison vector value with the latent feature comparison vector value to create a target comparison value for each of the digital assets, and reporting the digital assets having the highest target comparison values to the user or group of users.

A hybrid entity recognition system and accompanying method identify composite entities based on machine learning. An input sentence is received and is preprocessed to remove extraneous information, perform spelling correction, and perform grammar correction to generate a cleaned input sentence. A POS tagger tags parts of speech of the cleaned input sentence. A rules based entity recognizer module identifies first level entities in the cleaned input sentence. The cleaned input sentence is converted and translated into numeric vectors. Basic and composite entities are extracted from the cleaned input sentence using the numeric vectors.

A hybrid entity recognition system and accompanying method identify composite entities based on machine learning. An input sentence is received and is preprocessed to remove extraneous information, perform spelling correction, and perform grammar correction to generate a cleaned input sentence. A POS tagger tags parts of speech of the cleaned input sentence. A rules based entity recognizer module identifies first level entities in the cleaned input sentence. The cleaned input sentence is converted and translated into numeric vectors. Basic and composite entities are extracted from the cleaned input sentence using the numeric vectors.

A method and device for keyword extraction and a storage medium. The method includes receiving, at a terminal, an original document, acquiring, at the terminal, a candidate set by extracting at least one candidate phrase from the original document, acquiring, at the terminal, an association degree between the at least one candidate phrase in the candidate set and the original document, acquiring, at the terminal, a divergence degree of the at least one candidate phrase in the candidate set, and updating, at the terminal, a key phrase set of the original document by selecting the at least one candidate phrase from the candidate set as at least one key phrase based on the association degree and the divergence degree.

Methods and systems are provided for detecting a mental health condition. Structured and unstructured information is analyzed using natural language processing to extract information including clinical data values and medical concepts pertaining to a user. Reference medical information is evaluated using natural language processing to correlate medical data with mental health conditions. A classification for a mental health condition of the user is determined using a machine learning model and based on the extracted information and correlations, wherein the extracted information includes blood analysis for the user. The user is assigned to a segment of users based on the extracted information. A treatment for the mental health condition of the user is indicated based on the classification and the assigned segment of users.

The present application discloses a method, an apparatus, a device, and a storage medium for intention recommendation, which relates to the field of big data, artificial intelligence, intelligent search, information flow and deep learning technologies in the field of computer technologies. A specific implementation scheme includes: receiving an intention query request carrying an intention keyword and a user identification, determining a first recommendation list according to the intention keyword and a pre-configured intention repository, where the intention repository includes at least one tree-shaped intention set, and each tree-shaped intention set includes at least one graded intention, processing intentions in the first recommendation list by using intention strategy information corresponding to the user identification to obtain a target recommendation list and output it.

A digital attribution system is described to generate predictions of word attributions from subject data, e.g., titles, subject lines of emails, and so on. To do so, an attribution score is first generated by the digital attribution system that describe an amount to which respective words in the subject data cause performance of a corresponding outcome. The attribution scores are then used by the digital attribution system to generate representations for display in a user interface for respective words in the subject data and may also be used to generate attribution recommendations of changes to be made to the subject data.

A digital attribution system is described to generate predictions of word attributions from subject data, e.g., titles, subject lines of emails, and so on. To do so, an attribution score is first generated by the digital attribution system that describe an amount to which respective words in the subject data cause performance of a corresponding outcome. The attribution scores are then used by the digital attribution system to generate representations for display in a user interface for respective words in the subject data and may also be used to generate attribution recommendations of changes to be made to the subject data.