Natural language keyword tag extraction

An apparatus, comprising:

• a memory to store instructions;
• and
• processing circuitry, coupled with the memory, operable to execute the instructions, that when executed, cause the processing circuitry to:
• receive a natural language request for an automobile selection;
• in response to the received natural language request, apply a term frequency-inverse document frequency (TF-IDF) model to a plurality of terms of a data source comprising a corpus of reviews to compute a respective score for each of the terms, wherein the respective scores of a first subset of the plurality of terms exceed a threshold score;
• aggregating each term of the first subset of the plurality of term into a plurality of documents, wherein each document corresponds to a respective vehicular feature, of a plurality of vehicular features;
• rank the terms of the first subset based on the respective scores;
• generate, based on the aggregating and ranking of the terms of the first subset, a first plurality of tags related to a first automobile make and model of a plurality of automobile make and models, wherein each tag of the first plurality of tags is associated with a respective plurality of probability distributions generated by the TF-IDF model, each probability distribution associated with one of the plurality of vehicular features, the first plurality of tags further generated based on the probability distributions;
• generate, based on the respective scores of the terms of the first subset, a first co-occurrence matrix between the first plurality of tags and the plurality of automobile makes and models;
• generate, based on the first co-occurrence matrix, a second co-occurrence matrix between the first plurality of tags and a second plurality of tags;
• determine, based on the second co-occurrence matrix, a co-occurrence relationship between a first tag of the first plurality of tags and a second tag of the second plurality of tags, wherein the first tag and the second tag have scores exceeding a co-occurrence threshold;
• transmit a second suggestion including the first tag based on the co-occurrence relationship and the first tag having a higher score than the second tag;
• and
• update, by the TF-IDF model based on a selection of the first automobile make and model, the plurality of probability distributions of each tag of the first plurality of tags.

A non-transitory computer-readable storage medium storing computer-readable program code executable by a processor to:

• receive a natural language request for an automobile selection;
• apply a term frequency-inverse document frequency (TF-IDF) model to a plurality of terms in a data source comprising a corpus of reviews to compute a respective score for each of the terms, wherein the respective scores of a first subset of the plurality of terms exceed a threshold score;
• aggregate all terms of the first subset into a plurality of documents, wherein each document corresponds to a respective vehicular feature, of a plurality of vehicular features;
• configure a vehicle search interface with one or more filters based on the plurality of tags;
• rank the terms of the first subset based on the respective scores;
• generate, based on the aggregating and ranking of the terms of the first subset, a first plurality of tags related to a first automobile make and model of a plurality of automobile make and models, wherein each tag of the first plurality of tags is associated with a respective plurality of probability distributions generated by the TF-IDF model, each probability distribution associated with one of the plurality of vehicular features, the first plurality of tags further generated based on the probability distributions;
• transmit the plurality of tags in rank-order based on the scores of the ranked terms as a first suggestion for a particular automobile selection via the vehicle search interface;
• generate, based on the respective scores of the terms of the first subset, a first co-occurrence matrix between the first plurality of tags and the plurality of automobile makes and models;
• generate, based on the first co-occurrence matrix, a second co-occurrence matrix between the first plurality of tags and a second plurality of tags;
• determine, based on the second co-occurrence matrix, a co-occurrence relationship between a first tag of the first plurality of tags and a second tag of the second plurality of tags, wherein the first tag and the second tag have scores exceeding a co-occurrence threshold;
• transmit a second suggestion including the first tag based on the co-occurrence relationship and the first tag having a higher score than the second tag;
• and
• update, by the TF-IDF model based on a selection of the first automobile make and model via the vehicle search interface, the plurality of probability distributions of each tag of the first plurality of tags.