Patent•
Message processing method, device and system
07 Jul 2017-
TL;DR: In this paper, a message processing method consisting of obtaining spam messages marked by users, storing the spam message marked by the users to a spam message database, and reporting the spam messages in the spammed message database to a server, clustering users with similar spam message settings by the server, determining user groups with common spam message setting attributes based on a clustering result, and training customized spam message classifiers for the user groups.
Abstract: The invention discloses a message processing method. The method comprises the steps of obtaining spam messages marked by users; storing the spam messages marked by the users to a spam message database; and reporting the spam messages in the spam message database to a server, clustering users with similar spam message settings by the server, determining user groups with common spam message setting attributes based on a clustering result, and training customized spam message classifiers for the user groups with the common spam message setting attributes. The invention also discloses a message processing device and system. Through adoption of the technical scheme provided by the invention, spam message filtering can be carried out based on user preference, the use experience of the users is improved, and workloads of the server can be reduced.
References
More filters
Patent•
09 Nov 2009TL;DR: In this paper, a method and a system for recommending an ad (e.g., item) for a user is provided for recommending a product to a user based on user profiles and item attributes.
Abstract: A method and a system are provided for recommending an ad (e.g., item) for a user. In one example, the system constructs one or more user profiles. Each user profile is represented by a user feature set including user attributes. The system constructs one or more item profiles. Each item profile is represented by an item feature set including item attributes. The system receives historical item ratings given by one or more users. The system then generates one or more preference scores by modeling at least one relationship among the user profiles, the item profiles and the historical item ratings.
38 citations
Patent•
24 Dec 2014
TL;DR: In this article, a user behavior machine learning model training method and device is presented to solve the data sparseness problem without feature reduction and improves the accuracy of user behavior prediction, which includes collecting historical access data of a user; classifying and concentrating the historical access dataset of the user according to a characteristic set containing one or multiple dimensions, and acquiring a plurality of samples; calculating user behavior statistic information, including user's traffic quantity, corresponding to each sample; when the user traffic quantity corresponding to a current sample is smaller than a first threshold, calculating the distance between the current
Abstract: The invention discloses a user behavior machine learning model training method and device, solves the data sparseness problem without feature reduction and improves the accuracy of user behavior prediction. The method includes collecting historical access data of a user; classifying and concentrating the historical access data of the user according to a characteristic set containing one or multiple dimensions, and acquiring a plurality of samples; calculating user behavior statistic information, including user's traffic quantity, corresponding to each sample; when the user's traffic quantity corresponding to a current sample is smaller than a first threshold, calculating the distance between the current sample and the other samples; selecting the samples with the distances smaller than the threshold to serve as adjacent samples of the current sample; combining the user behavior statistic information of the current sample with the user behavior statistic information of the adjacent samples and generate new samples; utilizing new samples to train the pre-established machine learning model used for predicting the user behavior according to characteristic values of different dimensions of the characteristic set.
22 citations
Patent•
22 Apr 2009
TL;DR: In this paper, a method for filtering litter messages and a device thereof was proposed, which comprises: a litter message sending party number blacklist is acquired from a server; the blacklist comprises the telephone number complained by a user terminal in the server user group; the litter message is filtered based on the phone number of the acquired blacklist.
Abstract: The invention provides a method for filtering litter messages and a device thereof. The method comprises: a litter message sending party number blacklist is acquired from a server; the blacklist comprises the telephone number complained by a user terminal in the server user group; the litter message is filtered based on the telephone number of the acquired blacklist. By adopting the embodiment of the invention, the blacklist is generated based on the complaint of the user terminal in the user group, thus improving the success ration of litter message filtration.
19 citations
Patent•
10 Jun 2015
TL;DR: In this paper, a garbage message model training method was proposed, which consists of two steps: obtaining a message sample and performing feature extraction on the message sample to obtain a feature vector of the message samples.
Abstract: The invention discloses a garbage message model training method, a garbage message identifying method and a corresponding device. The garbage message model training method comprises the following steps: obtaining a message sample; performing feature extraction on the message sample to obtain a feature vector of the message sample; performing full-scale training on the message sample by adopting a learning monitoring mode to obtain a garbage message model, wherein the garbage message model, comprises conditional probability that the message including the feature corresponding to each feature in the feature vector is the garbage message. According to the technical scheme adopted by the invention, the garbage message model which is hidden but practically exists is found in a large number of message samples, so that the garbage information model obtained by training has capacity of precisely identifying. The garbage message identifying method disclosed by the invention can be used for precisely identifying whether different messages with the same feature are garbage messages or normal messages, so that the identifying accuracy rate is increased.
17 citations
Patent•
08 Jan 2014
TL;DR: In this article, the authors proposed a method for updating an existing junk short message classifier through the trained junk short messages classifier, which consists of receiving at least one short message and reporting information of the short messages reported by the terminal.
Abstract: The invention discloses a method, device, terminal, server and system for updating a classifier, and belongs to the technical field of short messages. The method comprises the steps of receiving at least one short message and reporting information of the short messages reported by the terminal, training a junk short message classifier according to the received short messages and the reporting information of the short messages, obtaining the trained junk short message classifier, and sending the trained junk short message classifier to the terminal. By means of the method for updating an existing junk short message classifier through the trained junk short message classifier, the problems that as a junk short message classifier is trained according to a large number of samples of the server, the junk short message classifier may not conform to use habits of every user, and misjudgments occur frequently are solved, and the effects of updating the existing junk short message classifier of the terminal according to the reporting information, lowering misjudgments on junk short messages and improving classification accuracy of the short massages are achieved.
6 citations