Abstract: The disclosure is directed to systems, methods, and computer storage media, for, among other things, employing nested model structures to enforce compliance, within a computational system, to at least one policy. One method includes receiving a digital record that encodes content. A plurality of models (e.g., integrated models and/or model droplets) is employed to analyze the records. The plurality of models is configured and arranged within a nested structure of a hierarchy of models. Each of the plurality of models analyzes at least a portion of the record. Based on the nested structure, the hierarchy combines the analysis from each of the plurality of models to determine that the content violates a policy of a system. In response to determining that the content violates the policy, at least one mitigation (or intervention) action are performed. The at least one mitigation action may alter subsequent transmissions of the record.

Abstract: The technology described herein determines whether a candidate text is in a requested class by using a generative model that may not be trained on the requested class. The present technology may use of a model trained primarily in an unsupervised mode, without requiring a large number of manual user-input examples of a label class. The may produce a semantically rich positive example of label text from a candidate text and label. Likewise, the technology may produce from the candidate text and the label a semantically rich negative example of label text. The labeling service makes use of a generative model to produce a generative result, which estimates the likelihood that the label properly applies to the candidate text. In another aspect, the technology is directed toward a method for obtaining a semantically rich example that is similar to a candidate text.

Abstract: The disclosure is directed to systems, methods, and computer storage media, for, among other things, employing nested model structures to enforce compliance, within a computational system, to at least one policy. One method includes receiving a digital record that encodes content. A plurality of models (e.g., integrated models and/or model droplets) is employed to analyze the records. The plurality of models is configured and arranged within a nested structure of a hierarchy of models. Each of the plurality of models analyzes at least a portion of the record. Based on the nested structure, the hierarchy combines the analysis from each of the plurality of models to determine that the content violates a policy of a system. In response to determining that the content violates the policy, at least one mitigation (or intervention) action are performed. The at least one mitigation action may alter subsequent transmissions of the record.

Abstract: The disclosure is directed to systems, methods, and computer storage media, for, among other things, generating, training, and tuning lexicon-based classifier models. The models may be employed in various compliance enforcement applications and/or tasks. The tradeoff between the model's false positive error rate (FPR) and the model's false negative rate (FNR) may be “tuned” via a balance parameter supplied by the user. The classifier model may classify content (e.g., text records) as either belonging to a “positive” class or a “negative” class. The positive class may be associated with non-compliance, while the negative class may be associated with compliance (or vice-versa). In some embodiments, the classifier model may be a probabilistic probability model that provides a probability (or degree of belief) that the content is associated with the positive and/or negative class.

Abstract: The technology described herein determines whether a candidate text is in a requested class by using a generative model that may not be trained on the requested class. The present technology may use of a model trained primarily in an unsupervised mode, without requiring a large number of manual user-input examples of a label class. The may produce a semantically rich positive example of label text from a candidate text and label. Likewise, the technology may produce from the candidate text and the label a semantically rich negative example of label text. The labeling service makes use of a generative model to produce a generative result, which estimates the likelihood that the label properly applies to the candidate text. In another aspect, the technology is directed toward a method for obtaining a semantically rich example that is similar to a candidate text.

Abstract: A method, computer program product, and computing system for associating one or more fashion products on a website with a user accessing the website. One or more recommendations may be provided to the user for fashion products based upon, at least in part, one or more fashion-ability scores representative of the one or more fashion products associated with the user on the website and one or more fashion-ability scores representative of the one or more fashion products on the website.

Abstract: Embodiments relate to an intelligent computer platform for computing visual similarity and identification of a product responsive to the computed similarity. A first product is selected, and one or more product attributes are identified. A multi-context similarity is dynamically assessed combining a vector map with a tensor representation, and applying a vector similarity algorithm against the map and representation to identify one or more similar objects. In response to a second product selection, the multi-context similarity is dynamically re-assessed based on proximity to identify and select a final product.

Abstract: A business forecasting tool utilizing metadata is provided. A processor receives one or more sets of business metrics. A processor receives a first metadata descriptor for a first set of business metrics of the one or more sets of business metrics. A processor receives a second metadata descriptor for a second set of business metrics of the one or more sets of business metrics. A processor prepares the first set of business metrics for prediction of a third set of business metrics based on, at least in part, the first metadata descriptor, where the first set and third set each correspond to a different time period. A processor generates a fourth set of business metrics based on, at least in part, the second metadata descriptor, where the second set and fourth set each correspond to a different time period.

Abstract: Approaches for automated fashion designing are described. A computer-implemented method for automated fashion designing includes: training, by a computer device, computer models using deep learning based computer vision; identifying, by the computer device, at least one gap using cognitively determined fashionability scores (F-scores); and creating, by the computer device, a new fashion design using the computer models and the at least one identified gap.

Abstract: A method, computer program product, and computing system are provided for identifying an advertising opportunity on a first website in response to a user accessing the first website. Information associated with the user accessing the first website may be received. One or more digital advertisements of one or more fashion products from the second website may be provided for rendering on the first website based upon, at least in part, one or more fashion-ability scores representative of the one or more fashion products on the second website and the information associated with the user accessing the first website.

Abstract: A first plurality of f-scores for a first plurality of products in an inventory are determined. The first plurality of f-scores are discretized to generate a plurality of groups, and a forecast is generated for each of the groups based on historical data associated with the first plurality of products. Projected gaps in the inventory are identified based on the forecasts. A second plurality of f-scores are determined for a second plurality of products, where each of the second plurality of products is not in the inventory. For each of the second plurality of products, a corresponding group in the plurality of groups is identified based on the second plurality of f-scores, an at least one product in the second plurality of products is selected to order, based on determining that the identified corresponding group aligns with at least one of the one or more projected gaps in inventory.

Abstract: Methods and systems for generating cognitive real-time pictorial summary scenes are disclosed. A method includes: obtaining, by a computing device, a document; training, by the computing device, computer models using natural language processing and deep learning based computer vision; and creating, by the computing device, a first pictorial summary scene that summarizes the document using the computer models.

Abstract: Embodiments relate to an intelligent computer platform to compute visual similarity across image objects. An object detection algorithm is utilized to identify image objects and to produce a tensor representation of the identified object. Multi-visual contextual similarity of the object is conducted to assess and determine related object images. A re-assessment of similarity is dynamically conducted in response to a product image selection. The re-assessment utilizes the tensor representations of the related object images, thereby conducting a mathematical assessment of similarity and object image identification. A final product is identified and selected based on the dynamic re-assessment and convergence on a directed outcome with minimal iterations of object interaction.

Abstract: Embodiments relate to an intelligent computer platform for computing visual similarity and identification of a product responsive to the computed similarity. A first product is selected, and one or more product attributes are identified. A multi-context similarity is dynamically assessed combining a vector map with a tensor representation, and applying a vector similarity algorithm against the map and representation to identify one or more similar objects. In response to a second product selection, the multi-context similarity is dynamically re-assessed based on proximity to identify and select a final product.

Abstract: A first plurality of f-scores for a first plurality of products in an inventory are determined. The first plurality of f-scores are discretized to generate a plurality of groups, and a forecast is generated for each of the groups based on historical data associated with the first plurality of products. Projected gaps in the inventory are identified based on the forecasts. A second plurality of f-scores are determined for a second plurality of products, where each of the second plurality of products is not in the inventory. For each of the second plurality of products, a corresponding group in the plurality of groups is identified based on the second plurality of f-scores, an at least one product in the second plurality of products is selected to order, based on determining that the identified corresponding group aligns with at least one of the one or more projected gaps in inventory.

Abstract: Embodiments relate to an intelligent computer platform to compute visual similarity across image objects. An object detection algorithm is utilized to identify image objects and to produce a tensor representation of the identified object. Multi-visual contextual similarity of the object is conducted to assess and determine related object images. A re-assessment of similarity is dynamically conducted in response to a product image selection. The re-assessment utilizes the tensor representations of the related object images, thereby conducting a mathematical assessment of similarity and object image identification. A final product is identified and selected based on the dynamic re-assessment and convergence on a directed outcome with minimal iterations of object interaction.

Abstract: As disclosed, f-scores can be generated for apparel items. Training images are identified, where each training image is associated with a corresponding set of tags including information about a plurality of attributes. A first convolutional neural network (CNN) is trained based on the plurality of training images and a first attribute. The first CNN is iteratively refined by, for each respective attribute, removing a set of neurons from the first CNN and retraining the first CNN based on the training images and the respective attribute. Upon determining that the first CNN has been trained based on each of the attributes, one or more CNNs are generated based on the first CNN. An image is received, where the image depicts an apparel item. The image is processed using the one or more CNNs, and an f-score for the apparel item is determined based on the output.

Abstract: As disclosed, f-scores can be generated for apparel items. Training images are identified, where each training image is associated with a corresponding set of tags including information about a plurality of attributes. A first convolutional neural network (CNN) is trained based on the plurality of training images and a first attribute. The first CNN is iteratively refined by, for each respective attribute, removing a set of neurons from the first CNN and retraining the first CNN based on the training images and the respective attribute. Upon determining that the first CNN has been trained based on each of the attributes, one or more CNNs are generated based on the first CNN. An image is received, where the image depicts an apparel item. The image is processed using the one or more CNNs, and an f-score for the apparel item is determined based on the output.

Abstract: Methods and systems for generating cognitive real-time pictorial summary scenes are disclosed. A method includes: obtaining, by a computing device, a document; training, by the computing device, computer models using natural language processing and deep learning based computer vision; and creating, by the computing device, a first pictorial summary scene that summarizes the document using the computer models.

Abstract: A method, computer program product, and computing system are provided for receiving a plurality of customer data for a plurality of retail spaces. A plurality of fashion-ability scores representative of a plurality of fashion products may be generated. An order for shipment of at least a subset of the plurality of fashion products to at least a subset of the plurality of retail spaces may be generated based upon, at least in part, the plurality of customer data and the plurality of fashion-ability scores.