An information processing system including a communication unit that acquires information related to an interaction between objects from a sensing device that detects the interaction between the objects, an emotion information database constructed by accumulating an evaluation value used when an emotion value of each object generated based on the information related to the interaction between the objects is calculated, a certification unit that certifies the sensing device and issues certification information to the sensing device, and an authentication unit that authenticates the information related to the interaction transmitted from the sensing device based on the certification information issued to the sensing device.

The present disclosure relates to improving recommendations for small shops on an ecommerce platform while maintaining accuracy for larger shops. The improvement is achieved by retraining a machine-learning recommendation model to reduce sample selection bias using a meta-learning process. The retraining process comprises identifying a sample subset of shops on the ecommerce platform, where the sample batch includes shops in both a small size category and shops in a large size category. The machine-learning model is then used to make shop-specific user-item interaction predictions for each shop in the sample batch. A shop-specific loss is calculated for each shop based on the model's predicted user-item interactions for the shop-specific training dataset and the actual user-item interactions in the shop-specific training dataset. A global loss is calculated based on each of the shop-specific losses, and the baseline model is updated to minimize the global loss.

The present disclosure relates to improving recommendations for small shops on an ecommerce platform while maintaining accuracy for larger shops. The improvement is achieved by retraining a machine-learning recommendation model to reduce sample selection bias using a meta-learning process. The retraining process comprises identifying a sample subset of shops on the ecommerce platform, and then creating shop-specific versions of the recommendation model for each of the shops in the subset. Each shop-specific model is created by optimizing the baseline model to predict user-item interactions in a first training dataset for the applicable shop. Each of the shop-specific models is then tested using a second training dataset for the shop. A loss is calculated for each shop-specific model based on the model's predicted user-item interactions and the actual user-item interactions in the second training dataset for the shop. A global loss is calculated based on each of the shop-specific losses, and the baseline model is updated to minimize the global loss. The model includes small and large-shop weight parameters that are applied to user-item interaction scores and that are learned during the re-training process.

The present disclosure relates to improving recommendations for small shops on an ecommerce platform while maintaining accuracy for larger shops. The improvement is achieved by retraining a machine-learning recommendation model to reduce sample selection bias using a meta-learning process. The retraining process comprises identifying a sample subset of shops on the ecommerce platform, and then creating shop-specific versions of the recommendation model for each of the shops in the subset. A global parameter adjustment is calculated for the global model based on minimizing losses associated with the shop-specific models and increasing the probability of items being recommended from small shops. The latter is achieved by introducing regularizer terms for small shops during the meta-learning process. The regularizer terms serve to increase the probability that an item from a small shop will be recommended, thereby countering the sample selection bias faced by small-shop items.

A system that analyzes a user's communications to select a promotion that is presented to the user. The analysis may occur in two stages: a first stage analyzes a single communication from a user to determine whether the user is a potential target for a promotion; for potential targets, a second stage analyzes a history of communications from the user to generate a user profile. The system may then select a promotion based on the profile. The profile may include a set of profile tags that are considerably more detailed and granular than traditional demographic data; tags may for example indicate user affiliations with groups or ideas (such as religions or political parties), or user life cycle stages. Using these rich, detailed user profile tags, the system may achieve promotion response rates far above those from traditional advertising, which relies on cookies or simple demographic categories.

Method and system for assessing the suitability of an entity using a proxy. A description of a behavior associated with a desirable audience is received. A proxy behavior estimated to be characteristic of the desirable audience is selected. The proxy behavior comprises the performance of proxy events related to the consumption of media received by an entity over a network, which can be found in an entity's consumption history. An entity can be assessed for inclusion in a proxy audience, by examining the entity's consumption history for proxy behaviors. A behavioral model is built using a training set comprising the proxy audience. By applying the behavioral model to the consumption history of a specified entity, the specified entity's suitability for selection can be determined. Advantageously, in an embodiment, the invention enables the use of behavioral modeling techniques even when the complete behavior of the desirable audience is not available.

An electronic marketplace provides a communication platform between consumer and sellers. Consumers create product wish lists and the wish lists are used as the basis for product advertising from the sellers. In addition, a consumer may invite friends to view the wish list and purchase particular products that are added to the wish list. Because wish lists are used as the basis for advertising, consumers can manage the types of product advertising that they see. Consumers have the ability to turn off individual product advertisements by removing that product from the wish list. Sellers utilize reverse bidding to compete for advertising space with each consumer and friend. One particularly advantageous feature of the embodiment is that wish lists serve as nameless and untraceable proxies for each member by keeping personal information removed from the sellers.

A method for presenting pre-approved and pre-underwritten offers to a customer may include: receiving targeting criteria based on at least one of current accounts with the financial institution, assets, creditworthiness, and credit risk for an offer for a financial product; identifying a target population of customers for the offer by applying the targeting criteria to a population of customers; reviewing each customer in the target population for underwriting for the financial product based on inferred income for each customer and accounts that each customer has with the financial institution, wherein the underwriting is performed before the financial product is offered; determining a channel to present the offer to one of the customers that passed underwriting; communicating the offer to the customer over the selected channel; and providing an accepted offer to a fulfilment engine, wherein the fulfilment engine initiates an account opening for the financial product.

An online system accesses a model trained based on a topic associated with a set of content items and the content of the set of content items. The online system applies the model to predict a probability that each of multiple content items is associated with the topic based on its content and identifies (a) content item(s) associated with at least a threshold probability. The online system retrieves information describing user engagement with the identified content item(s) and determines a value of a performance metric for the topic based on this information. If the value is at least a threshold value and the online system receives content from an entity describing an item associated with the topic, the online system communicates a recommendation to the entity to create a content item describing the item and to add a tag associated with the item upon determining an opportunity to do so.

According to examples, a system for automatically optimizing thresholds of content processing models that select content for presentation to users may include a processor and a memory storing instructions. The processor, when executing the instructions, may cause the system to select a subset of the content processing models for a content policy grouping. The subset of content processing models comprises models selected from a plurality of content processing models based on content rejection rates and models that are selected based on corresponding model probabilities. The system may further obtain an optimized threshold for each model of the subset of content processing models based on an iterative global optimization technique. The system may thereby facilitate automatic selection or rejection of the content pieces for presentation to users on an online system based on the policies associated with corresponding content policy grouping by employing the subset of content processing models with the optimized thresholds.