Abstract: This disclosure relates to systems, methods, and computer readable media for performing multi-format, multi-protocol message threading in a way that is most beneficial for the individual user. Users desire a system that will provide for ease of message threading by “stitching” together related communications in a manner that is seamless from the user's perspective. Such stitching together of communications across multiple formats and protocols may occur, e.g., by: 1) direct user action in a centralized communications application (e.g., by a user clicking ‘Reply’ on a particular message); 2) using semantic matching (or other search-style message association techniques); 3) element-matching (e.g., matching on subject lines or senders/recipients/similar quoted text, etc.); and 4) “state-matching” (e.g., associating messages if they are specifically tagged as being related to another message, sender, etc. by a third-party service, e.g., a webmail provider or Instant Messaging (IM) service).

Abstract: An approach is provided in which a fault-injecting system injects a natural language fault into a first text segment to produce a second text segment that are both written in a natural language. The fault-injecting system receives a third text segment from a reviewer that includes at least one correction to the second text segment. The fault-injecting system compares the third text segment against the first text segment and generates an efficacy score. The efficacy score indicates whether the correction in the third text segment corrects the natural language fault. In turn, the fault-injecting system sends the efficacy score to an author of the first text segment.

Abstract: Embodiments of the invention include a context sensitive adaptive digital assistant for personalized interaction. Embodiments of the invention also include a spoken genome for characterization and analysis of human voice. Aspects of the invention include selecting a starter vocabulary, receiving voice communications from a user, and modifying the starter vocabulary to generate a personalized lexicon. Aspects of the invention also include analyzing and categorizing human voice according to a plurality of characteristics, and creating a spoken genome database.

Abstract: A method for identifying actions of an electric equipment is provided. The method includes: measuring a utility power signal of the electric equipment during a period of time; analyzing a waveform characteristic of the utility power signal in the period of time to acquire a set of characteristic values, and identifying a current control procedure corresponding to the utility power signal from a set of control procedures according to the set of characteristic values, the current control procedure including one or more action stages whose first action stage corresponds to a first action parameter; determining whether a deviation between the waveform characteristic of the utility power signal corresponding to the first action stage and the first action parameter is within a tolerance range; triggering a warning event when the deviation between the waveform characteristic of the utility power signal and the first action parameter is outside the tolerance range.

Abstract: Systems and methods for measuring brand uplift in a computer networked environment are provided. A data processing system can assign a first set of device identifiers to an experimental arm and a second set of device identifiers to a control arm. Computing devices associated with the experimental arm receive third-party content items promoting a brand for which a brand uplift experiment is performed, while computing devices associated with the control arm do not receive third-party content items promoting the brand. At the end of the experiment, a brand uplift metric value is computed by analyzing search query logs of the computing devices. The brand uplift metric value is then provided to an advertiser for display.

Abstract: Techniques are disclosed for adjusting a ranking of information content presented to a user based on voice-of-customer feedback. In one embodiment, a user may provide feedback on information content presented to the user. Such feedback may be evaluated to identify at least one topic referenced in the received feedback. If an application determines that the at least one topic is related to topics of the information content, the application determines sentiment regarding the information content based on the feedback, and adjusts a ranking of the information content based on the determined sentiment.

Abstract: A ladder-program managing device includes: a database that has a ladder-program-storing region in which ladder program for individually controlling a plurality of pieces of machinery are stored, and that contains ladder-program-history information storing, in association with each other, identification information for identifying contents of alterations that have already been applied to the plurality of pieces of machinery and correction programs required to perform the alterations; and a database managing portion that manages the database, wherein the database managing portion is provided with an identification-information-searching portion that searches for, in the ladder-program-history information, identification information corresponding to an alteration content to be applied to an alteration-target machinery, and a merging portion that identifies the correction program corresponding to the searched identification information in the ladder-program-history information, and that merges the correction progr

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, modifying queries based on non-textual content. In one aspect, a method includes receiving, from a user device, a query including a plurality of terms; determining active non-textual data displayed in an application environment on the user device; determining, from the non-textual textual data, modification data for the query; generating a set of modified queries based on the query and the modification parameters; scoring the modified queries according to one or more scoring criteria; selecting one of the modified queries based on the scoring; and providing, to the user device, search results responsive to the selected modified query.

Abstract: Various technologies related to generating and applying content retrieval rules are described herein. A content retrieval rule maps a combination of a query and a context to one of a query reformulation or content. The content retrieval rule is learned from search logs of a search engine, and is applied when the query having the context is received at the search engine.

Abstract: Automatically creating and modifying a search engine for a website. User input may be received specifying an address of a website. A search engine may be automatically created for the website based on the user input. Webpages of the website may specify a plurality of tags specifying custom attributes of the webpages. During creation of the search engine, these custom attributes may be incorporated into the search engine index. Additional user input may be received customizing the search engine for various search engine contexts, e.g., based on the custom attributes of the webpages. Search engine results for the website may be based on various ranking functions, potentially including social impact of webpages of the website.

Abstract: A method including selectively aggregating a first set of filters for an item results list. The first set of filters can include multiple filter types, including category filters, facet filters, and price filters. The method also can include, for each filter in the first set of filters and each item in the item results list, determining user engagement statistics for the item when the filter has been applied. The method additionally can include generating a filter score for each filter in the first set of filters. The method further can include selecting a second set of filters from the first set of filters based on the filter scores of the filters in the second set of filters being above a threshold filter score. The method additionally can include applying space-constraint rules to the second set of filters to limit a quantity of filters in the second set of filters based on a ranking of the filters in the second set of filters and based on a screen size of a user device.

Abstract: The present disclosure is directed toward systems, methods, and non-transitory computer readable media that use multi-point optimization for delivery of digital content by way of a digital content distribution platform. In particular, one or more embodiments described herein receive a content item from a content provider to be displayed to users of the platform. The embodiments optimize delivery to obtain a first target event and determine metrics that delivery of the content item is expected to satisfy. If the actual metrics of delivery fail to satisfy the expected metrics, delivery is re-optimized to obtain either the first target event or a second target event.

Abstract: Techniques are disclosed for augmenting an interactive communication session with contextually relevant information. A computing device, such as a smartphone or tablet computer, can be configured to provide an augmented chat application configured to create, send, receive and display a conversation, and further configured to locate, retrieve and display, via a graphical user interface (GUI), other information that is contextually relevant to the conversation. The contextual information may be shared with one or more other users who are participating in the conversation either automatically or in response to a user input. The contextual information can be displayed or otherwise presented to the user explicitly (e.g., the actual content of the contextual information can be displayed via the GUI) or implicitly (e.g., a referent to the contextual information, such as a hyperlink or other cue, can be displayed via the GUI).

Abstract: In one example, a network management system (NMS) and method models services in a network. The NMS identifies externally mapped attributes in a vendor neutral yang model of a network service, assigns an attribute type to one or more of the externally mapped attributes, wherein assigning includes determining, by a learning system, the attribute type of one or more of the externally mapped attributes based on an analysis of attribute types assigned to the externally mapped attributes of other network services, constructs a separate dependency graph for each of the externally mapped attributes and builds a separate hidden service model for each of the externally mapped attributes, wherein the externally mapped attribute for each hidden service model is a merge attribute for that hidden service model.

Abstract: An online system ranks topic-groups for users and presents content items in topic-based feeds. A topic group corresponds to one or more topic(s) and can be used to generate a feed for presenting the content items related to the topic(s). For a particular user, the topic groups are ranked according to the likelihood of the user interacting with content items included in the topic groups. The topic groups are ranked using information of the users and/or users' historical interaction data such as click-based interaction data, post-based interaction data, or engagement-based interaction data. The online system generates and provides a user interface for presenting the topic groups to the client device. Content items that are related to the topic(s) corresponding to the topic group are presented in each topic-based feed such that the user can switch between different topic-based feeds.

Abstract: A system and method for managing display of database records are provided. The method includes determining whether a pointer is positioned over a data cell of a plurality of data cells of a graphical user interface; in response to receiving an actuation signal related to the pointer while the pointer is determined to be over the data cell, automatically modifying search criteria applied to database data according to the data of the data cell; and automatically outputting the database data filtered by the modified search criteria to the graphical user interface.

Abstract: A device may receive first data items and second data items that may have been output by sensor devices. The device may add, to a node in a data structure, at least some of the first data items. The device may divide the node to create subnodes that correspond to subregions of a geographic region. The device may add, to at least one of the subnodes, at least some of the second data items. The device may generate a probability of a first relationship, between the first data items and the second data items, based on determining subnodes that include a first data item and determining subnodes that exhibit a predefined second relationship with a second data item. The device may send, to another device, the probability of the first relationship to support location-based services.

Abstract: Techniques for presenting data visualizations are described. Consistent with some embodiments of the present invention, a data visualization system includes an approximate query engine to derive approximate query results and present approximate data visualizations, in interactive time. The data visualization system also includes a precise query engine, thereby enabling a data analyst to selectively request that a precise query result be generated for a specific query, so that the data analyst can confirm a data observation made when viewing the approximate visualization for that specific query. The precise visualization is presented in a manner that allows the data analyst to view, and thus confirm or disprove, a data observation recorded when viewing the approximate visualization, and also compare the precise visualization with the previously generated and presented approximate visualization.

Abstract: In a deduplicated microservices-based storage system and method running in a cloud includes frontend and backend microservice nodes that process requests from a plurality of clients and interface with a master node and one or more worker nodes connected as a network for deduplicating, storing and retrieving data from object storage, a worker node receives queries and fingerprints for data segments of a data similarity group to be deduplicated, obtains and returns the similarity group from its cache or from object storage using a key. The system compares the received fingerprints to stored fingerprints to determine non-duplicates for deduplication. The worker node updates similarity groups in its cache and in a write-ahead log with a reference to a compression region of non-duplicate segments, and periodically writes modified objects from its cache to object storage.

Abstract: A machine may be configured to generate a digital content feed based on at least explicit feedback from a member of a Social Networking Service (SNS). For example, the machine generates explicit input data describing digital content preferences of a member of the SNS based on a communication including explicit feedback data. The communication is received from a client device associated with the member. The machine accesses feature data pertaining to one or more items of digital content determined to be relevant to the member. The feature data describes one or more characteristics associated with the one or more items. The machine generates a feed of items of digital content for the member based on the explicit input data and the feature data. The machine causes a presentation of the feed of the items of digital content in a user interface of the client device associated with the user.

Abstract: Presented are systems and methods that estimate the strength of a relationship between elements gathered from online and/or offline information sources by estimating the trustworthiness both of the gathered data and the information sources from which the data originates. In one exemplary application, the relatedness between co-occurring symptom and disease terms collected from information sources, such as health-related online databases, is iteratively evaluated based on the trustworthiness of symptom-disease pairings and the trustworthiness of the information sources themselves. In various embodiments of the present disclosure, an objective function is used to extract a knowledge base that aids in identifying a potential relationship between a set of given symptoms provided by a user of an online healthcare service and co-occurring disease terms, such that a likely disease may be inferred from the set of symptoms.

Abstract: In one embodiment, a method includes accessing a first set of scored results received from a client system of a first user, the first set of scored results including: one or more results, each having a rank with respect to the other results, generated by a first search algorithm, and one or more scores inputted by the first user corresponding to the results. The method includes calculating a discounted cumulative gain for each scored result based on the rank and score of the result. The method includes generating a second search algorithm based on the calculated gains and the first search algorithm, the second search algorithm including a modification of the first search algorithm to improve the ranking of results, and, if a quality metric of the second search algorithm is higher than a quality metric of the first, substituting the second search algorithm for the first.

Abstract: This invention solves the long-standing problem in Machine Learning of training a neural network on a spike-based neuromorphic computer. The preferred embodiment of the invention describes an algorithm for training a Restricted Boltzmann Machine (RBM) neural network, but the invention applies equally to training neural networks in the general class of Markov Random Fields. The standard CD algorithm for training an RBM on a general-purpose computer is unsuitable for implementation on a neuromorphic computer, as it requires the communication of real-valued parameter values between neurons, and/or shared memory access by neurons to stored parameter values. By employing the invention described, these requirements are eliminated, thus providing a training algorithm which can be implemented efficiently on a spike-based, distributed processor and memory, neuromorphic computer system.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for composition aware image querying. A system may receive user input identifying a search query for content from a client device, where the user input indicates one or more queries assigned to one or more regions of a search input page. The system may generate a query vector for each query using a computer-operated neural language model. The system may compare the query vector to an indexed vector for each region of an image. The system may determine a listing of composition-specific images from a collection of images based on the comparison. The system may determine a ranking for each image in the listing of composition-specific images, and provide search results responsive to the search query to the client device. The search results may include a prioritized listing of the composition-specific images based on the determined ranking.

Abstract: In various embodiments, a method and an apparatus for extracting topic sentences of webpages are provided. The method comprises: obtaining candidate webpages, and a pre-built machine learning model, each candidate webpage contains multiple preselected candidate topic sentences, and each candidate topic sentence includes several word segments; determining word feature values that indicate importance levels of the word segments in each candidate webpage respectively, and inputting the word feature values to the machine learning model to obtain an importance value for each word segment; for each candidate webpage, determining a partial order value for each candidate topic sentence according to the importance values of the word segments included in the candidate topic sentence; and for each candidate webpage, selecting one of the plurality of candidate topic sentences that is associated with a partial order value larger than a preset threshold value as a target topic sentence of the candidate webpage.

Abstract: Some embodiments of the present invention include a method for identifying unique and duplicate queries from a group of queries associated with a database system. The method includes forming, by a server computing system, two or more subgroups of queries from a group of queries based on a characteristic of the queries, the group of queries associated with a first database system; determining, by the server computing system, unique and duplicate queries in each of the subgroups based on a similarity threshold; and migrating, by the server computing system, the unique queries from each of the subgroups to a second database system.

Abstract: An application recommendation method and a recommendation apparatus are disclosed. A method comprises: acquiring a popularity value and a security risk value of each application in multiple applications; and ranking the multiple applications according to the popularity value and the security risk value. In a technical solution of embodiments of the present application, multiple applications are ranked according to popularities and security risk vales of the applications to preferentially recommend, to a user, an application that better meets popularity and security preferences of a user, to cause the user being protected in terms of security and/or privacy while enjoying the application having abundant functions.

Abstract: A system for extracting attributes can analyze text from data sources, extract n-grams from the text as candidate attribute and service/product pairs, prompt a human operator to rate the suitability of the candidate attribute and service/product pairs, and, based on the ratings, add the candidate attribute and service/product pairs to an attribute dictionary. In embodiments, an attribute extraction system includes an n-gram extractor, an attribute selector interface, and a dictionary builder. Data sources may include product titles, category descriptions, product descriptions, and like data from one or more product databases. In embodiments, the attribute dictionary is analyzed to determine canonical names for products or services and name variants for the products or services.

Abstract: There is provided a method for determining a search response to a search query associated with a user. The method may be executable at a server. The method comprises responsive to a receipt of the search query, determining the most relevant document to the search query; determining a likelihood parameter indicative of how likely the most relevant document is to satisfy the search query; in response to the likelihood being above a threshold, displaying exclusively the most relevant document; in response to the likelihood being below the threshold, displaying the general SERP including the most relevant document and other documents.

Abstract: Various aspects of the subject technology relate to systems, methods, and machine-readable media for identification of a salient portion of an image. A system may receive user input identifying a search query for content from a client device. The system may determine a listing of images responsive to the search query from an image collection. The system may obtain one or more image crops for at least one image of the listing of images based on a saliency map of the at least one image. In one or more implementations, each of the one or more image crops indicates a salient region of a corresponding image. The system may provide a set of search results responsive to the search query to the client device. In one or more implementations, the set of search results includes the obtained one or more image crops in a prioritized listing of image crops.

Abstract: Various technologies related to generating and applying content retrieval rules are described herein. A content retrieval rule maps a combination of a query and a context to one of a query reformulation or content. The content retrieval rule is learned from search logs of a search engine, and is applied when the query having the context is received at the search engine.

Abstract: There is disclosed a computer-implemented method of generating a content recommendation for a user of an electronic device, the method executable by a recommendation, the content recommendation being associated with a content item available at one of a plurality of network resources accessible via the communication network. The method comprises: executing a first machine learning algorithm module in order to determine a sub-set of recommended content sources from a plurality of possible content sources that is based on at least some of a first sub-set of user-specific content sources and a generated second sub-set of user-non-specific content sources; analyzing the sub-set of recommended content sources to select a plurality of potentially-recommendable content items; executing a second machine learning algorithm module in order to select, from the plurality of potentially-recommendable content items, at least one recommended content item; the selection being made on the basis of a user-profile-vector.

Abstract: Aspects of the subject disclosure are directed towards increasing a community of interest group users of an interest group, e.g., a network site having a feed of posts generally related to a particular topic. Many newly created interest groups do not have enough content to keep users interested and attract new users. The technology described herein automatically obtains content that backfills such an interest group with additional related content. The backfilled content may be found by keywords search and/or by crawling general public data, data from specific sites, historical conversation data, and/or more widely scoped data. Also described is creating a new interest group and seeding the group with content based upon backfilling technology.

Abstract: Methods and devices for identifying available applications for use with hardware of a computer device may include generating an application request with at least one device identification and a hardware identification to determine available applications. The methods and devices may include transmitting the application request to a remote application store. The methods and devices may include receiving, from the remote application store, an available application list with one or more applications having application relationship metadata that identifies compatible usage with the at least one hardware identification. The methods and devices may include presenting the available application list on a display of the computer device.

Abstract: The disclosed embodiments illustrate methods and systems for predicting requirements of a user for resources. The method includes transforming a message posted by the user into a first message vector. The method further includes categorizing one or more first message vectors into one or more categories. The method further includes transforming each of the categorized first message vectors into one or more second message vectors using a wavelet transform technique. The method further includes determining, for each of the categorized first message vectors, a first score based on at least a probability distribution of one or more coefficients associated with each associated feature. The method further includes selecting a predefined number of features based on at least the first score. The method further includes training one or more classifiers on the selected predefined number of features to identify at least the one or more needs of the user.

Abstract: An optimized fact checking system analyzes and determines the factual accuracy of information and/or characterizes the information by comparing the information with source information. The optimized fact checking system automatically monitors information, processes the information, fact checks the information in an optimized manner and/or provides a status of the information. In some embodiments, the optimized fact checking system generates, aggregates, and/or summarizes content.

Abstract: A method for sorting a data table is provided. The method may include providing a plurality of attribute value information for each data block of the data table. The method may also include receiving a query requiring a sorting on the first attribute of the data table. The method may further include determining a plurality of sequences of a plurality of data blocks having disjoint value ranges of the first attribute based on the provided plurality of attribute value information. The method may also include, for each determined sequence of the plurality of data blocks, reading a plurality of data, sorting the read plurality of data from each data block, and concatenating the sorted plurality of data from the plurality of data blocks within the determined sequence, thereby providing a sorted plurality of sequences. The method may further include merging the sorted plurality of sequences.

Abstract: Methods, systems, and apparatus include computer programs encoded on a computer-readable storage medium, including a method for providing content. A search query is received. Search results responsive to the query are identified, including identifying a first search result in a top set of search results that is associated with a brand. Based at least in part on the query, one or more eligible content items are identified for delivery along with the search results responsive to the query. A determination is made as to when at least one of the eligible content items is associated with a same brand as the brand associated with the first search result. The first search result and one of the determined at least one eligible content items are combined into a combined content item and providing the combined content item as a search result responsive to the request.

Abstract: There is provided systems and method for content inversion for use in user searches and product recommendations. The methods include receiving a first content, wherein the first content includes a first sentence having at least one first sentiment expression, creating a syntactic parse tree of the first sentence, identifying a first sentiment in the at least one first sentiment expression, wherein the first sentiment corresponds to a first polarity, determining a first needs expression corresponding to the first sentiment, wherein the first needs expression includes a first subject. The method may further include creating a sub-tree corresponding to the first needs expression, and grouping the content with other content having similar sub-trees.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting content for user devices specific to a particular user. The aspects generally relate to providing related content items to a particular user across multiple different user devices, such as mobile devices, tablet computers, desktop computers, and televisions. The selection of the content items for a particular user and the device to which the content items are to be sent are based one or more of the following factors: user history data, subject matter category duration, frequency duration adjustment, behavioral duration adjustment, and device prioritization.

Abstract: A query processor determines if an initial query results in a query system having a search engine coupled to a database to identify a limited maximum number of records deliverable by the query system. This limitation is indicative of an incomplete result because records are includable in the query may not be included in the result. In response the query processor generates a plurality of queries that cause the query system to generate results that identify less than the maximum number. The results of the plurality of queries are combined as a response to the initial query. Duplicate records caused by any overlap in the plurality of queries may be eliminated from the combination. Determining a median value of a query range that causes the query system to deliver the maximum number of records may be utilized in the generation of the plurality of queries.

Abstract: The present invention provides a digital communications system for receiving and integrating multiple service communication feeds into an overall composite deliverable by way of an enhanced and interactive user interface. The system receives peer identification and scoring communications signals from multiple services and based on known handshake configurations links peer identification data to dimensions used to generate an integrated peer identification score and list. A graphical user interface allows users to modify the dimension aggregator function to obtain customizable sets of results based on services communications feeds.

Abstract: A new approach is proposed to support two variants of the approach to effectively generate reason codes for an ensemble model. Both variants involve treating the ensemble model as a black box, identifying trivial values for input variables to the ensemble model, replacing each of the input variables to the ensemble model with its corresponding trivial values and then evaluating the impact on a score of the model after the trivial value replacement. The evaluation result of the impact is then used to generate the reason codes for the ensemble model. Specifically, the first variant is configured to perform one round of replacement, wherein variables with top drops in the score will be the reason codes. The second variant is configured to perform multiple rounds of replacement, which in each round, keeps the identified reason codes variables replaced with trivial values and analyzes incremental drops of replacing remaining variables.

Abstract: The present teaching relates to providing a query suggestion. In one example, a request is received for query suggestions with respect to a query prefix input by a user. A plurality of query suggestions is determined based on the query prefix and a preceding query input by the user. A degree of popularity of the preceding query is determined. One or more query suggestions are selected from the plurality of query suggestions based on the degree of popularity of the preceding query. The one or more query suggestions are provided as a response to the request.

Abstract: A system and method for calculating quality score for digital content are provided. In example embodiments, a first graph is generated comprising a user node and an article node, the user node corresponds to a user and the article node corresponds to an article. An edge is generated between the user node and the article node in the first graph based on a first action. A second graph is generated comprising the user node and the article node. An edge is generated between the user node and the article node in the second graph based on a second action type. A first authority score is calculated for the article node within the first graph. A second authority score is calculated for the article node within the second graph. A quality score is calculated for the article.

Abstract: A method for determining a pixel value of a texture pixel associated with a three-dimensional scan of an object includes prioritizing a sequence of image frames in a queue based on one or more prioritization parameters. The method also includes selecting a first image frame from the queue. The method also includes determining a pixel value of the particular texture pixel in the first image frame. The method further includes selecting a second image frame from the queue. The second image frame has a higher priority than the first image frame based on the one or more prioritization parameters. The method also includes modifying the pixel value of the particular texture pixel based on a pixel value of the particular texture pixel in the second image frame to generate a modified pixel value of the particular texture pixel.

Abstract: A method and system for improving a router based search query is provided. The method includes identifying a Web page retrieved during a Web query received from a network router device and analyzing historical Web pages retrieved during historical search queries associated with a search engine router and the network router device. The Web page is ranked with respect to the historical Web pages and the Web page is tagged with a first domain comprising a topic associated with the Web page. An address for the network router device is tagged with the first domain and additional domains comprising additional topics associated with additional Web pages accessed via the network router device. In response, the Web page is re-ranked with respect to the historical Web pages.

Abstract: Media files may be prioritized based on user behavior data and social data associated with individual media files. A process of prioritizing media files may include accessing user behavior data associated with individual media files in a set of media files that are stored on a client device of a user or in a cloud-based storage location associated with the user, accessing social data associated with one or more media files of the set of media files that are determined to have been shared on a media file sharing service, and ranking the set of media files based at least in part on the user behavior data and the social data.

Abstract: Disclosed are systems, apparatus, and methods for integrating an information feed. In various implementations, an identity of a user may be determined based on authentication information, where the authentication information identifies a user profile. In some implementations, profile information is identified based on the determined identity, where the profile information identifies one or more entities tracked using one or more information feeds associated with the user profile, and where the one or more information feeds comprises one or more feed items stored in a database system. In various implementations, the identified profile information is associated with a user account provided by a network communications application.

Abstract: Methods and systems for determining the preferences of hosts offering accommodations are disclosed. In one embodiment, an online booking system models the preferences of hosts based on statistical relationships between features of previously received accommodation reservation requests and the acceptance of those reservation requests by the hosts. In particular, the system classifies reservation requests based on several features—a reservation request either possesses a feature or does not possess a feature. The preference of a host for a particular request feature is modeled based on the relationship between the reservation requests that possess the feature and the reservation requests that are accepted by the host.