103results about How to "Improve relevance" patented technology

A method for presenting dynamic advertising comprises providing an advertising presentation device; regularly receiving electronic context data related to the advertising presentation device; on the basis of the context data, automatically generating context-based advertising; and causing the advertising presentation device to present the context-based advertising. The method is useful in particular for mobile advertising presentation devices, as it enables advertising to be dynamically generated based on regularly received context data such as presentation device location, date/time, user input at the advertising presentation device. Increased relevance of the advertisement to the observer is therefore provided. A system for presenting dynamic advertisements comprises an advertising presentation device; a controller in communications with the advertising presentation device, the controller for: regularly receiving electronic context data related to the advertising presentation device; automatically generating context-based advertising on the basis of the context data; and causing the advertising presentation device to present the context-based advertising.

An advertisement can be presented by associating a label with an advertisement, receiving input from a user, wherein the input comprises one or more labels, identifying a match between the label associated with the advertisement and the input, and presenting the advertisement to the user. A bid further can be received for presenting the advertisement in response to input from a user that includes the associated label and an advertisement presentation order can be determined based on one or more of the bid, a click-through rate of the advertisement, and a conversion rate of the advertisement. Additionally, a negative label can be associated with the advertisement. Further, an additional input comprising one or more labels can be received from the user, a match can be identified between the negative label and the received additional input, and it can be determined not to present the advertisement based on the identified match.

The invention relates to the field of communication and provides a method and a system for carrying out association on users and friends thereof in a network community. The method comprises the following steps: A, extracting characteristic elements from personnel information of the users; B, searching resources relevant to the characteristic elements in the network community; C, confirming the friends to be recommended according to the searched resources and associating the friends to the users. The system comprises an extracting unit, a searching unit and an associating unit. The invention enhances the association between the users and the recommended friends, and improves the communication between the users and the friends in the network community.

Two-way communication between a mobile device and a server is used to provide increased relevance of traffic information at lower bandwidth. The complexity, amount of storage and processing power required of the mobile navigation device are minimized, since the server performs much of the required computing. A transactional billing model can be used to bill the driver based on usage or consumption of traffic information. A driver selects a destination and routing preferences on a mobile device, and the local system generates a routing. The mobile device sends the planned route and preferences to a server. The server uses real-time traffic flow and incident information to determine an estimated driving time for the planned route. The server may also generate alternative routes, with an estimated time for each alternative route. The traffic and time information for the planned and alternative routes are then transmitted back to the mobile device.

The invention discloses a system enabling recommended web site information to be displayed in a browser address bar. The system comprises a browser and a net web site server. The net web site server is suitable for receiving user information and/or input key word information sent by the browser, searching a net to recommend a web site list according to the user information and/or the input key word information, and sending back the net recommended web site list to the browser. The browser comprises an acquiring module, a sending module, a receiving module and a displaying module. The acquiring module is suitable for acquiring the key word information input in the address bar by a browser register user. The sending module is suitable for sending the user information of the browser register user and/or the input key word information to the net web site server. The receiving module is suitable for receiving the net recommended web site list returned by the net web site server based on the user information and/or the input key word information. The displaying module is suitable for displaying the net recommended web site information of the net recommended web site list in a browser address bar pull-down list.

The invention provides method and device for evaluating video quality. The method for evaluating video quality comprises the following steps: obtaining subjective space domain distortion by the objective space domain distortion of a detected video sequence relative to a reference video sequence; obtaining subjective time domain change by the subjective space domain distortion and evaluating the video quality according to the subjective time domain change. The device for evaluating video quality comprises a subjective space domain distortion generating module, a subjective time domain change generating module and a video quality generating module. A video quality evaluating result of the invention has higher relevance with an evaluating result of a method for evaluating subjective video quality.

The invention discloses a feedback method for channel state information. The feedback method includes the steps that UE receives a mode which is configured by a base station and used for reporting the channel station information (CSI); the UE measures a base station signal and interference, determines a rank, pre-coding matrixes and CQI used for downlink transmission under the current channel condition and obtains PMIs for instructing the pre-coding matrix according to the pre-coding matrixes; the UE reports the rank, the PMIs and the CQI in the uplink direction. A codebook used for feeding back the CSI is a sub-codebook of a standard codebook, and the higher the correlation between any two codes in the codebook is, the smaller the code distance between the two corresponding PMIs is. The feedback method can be applied to improving the transmission performance of a system.

The present invention provides a method for diagnosing, detecting and monitoring brain function, especially neurological diseases and disorders. This invention examines the output of a neurological monitoring device such as an electroencephalography (EEG) recording. The EEG recording is often taken while a person is engaged in a specific neurological task such as delayed recognition. This invention provides for two methods for the diagnosis, detection and brain monitoring based on the EEG recording. The first is the use of the person as their own baseline for comparison. The efficacy of a person's brain function is measured by comparing a portion of their EEG recording with a different portion. Each of these portions is taken from the same EEG recording of a single neurological task performance. The second method is the minimal use of monitoring device output, such as an EEG recording, in a manner congruent with the neurological task being performed by the person. For example, to test a person's delayed recognition memory; a person would first be required to perform a delayed recognition memory task. Then, the EEG recording for two electrodes, P3 and P4 would be examined for the first 150 milliseconds after recognition memory stimulus onset. Then the EEG recording for two other electrodes, T7FPp1 and T8Fp2 would be examined for the next 150 milliseconds. Since this choice of electrodes and times is congruent with the neurological task of delayed recognition, the data is highly relevant to the monitoring of the person's delayed recognition memory. This method, combined with using a person as their own baseline allows this invention to provide high accuracy in the detection, diagnosis and monitoring of brain function, especially neurological diseases and disorders.

The invention provides a method for constructing a deep convolution neural network model, belonging to the field of mode recognition and machine learning. The method comprises the steps of (1) initializing a convolution neural network model, (2) carrying out end-to-end global extension learning on a network until the average error of a convolution neural network system reaches a preset expected value, (3) after global extension, using a cross validation sample to evaluate network performance, and carrying out local extension learning on the network if a recognition rate does not reach the expected value, (4) adding a new incremental end-to-end extension learning branch, realizing the incremental expansion learning of a network structure, and finally realizing the model construction of the deep convolution neural network. According to the method, a neural element can be added according to needs according to a condition of participating in training samples, the extension expansion and incremental expansion of the network structure are realized, the relevance between a data sample and the network model is enhanced, and the network structure adaptive incremental learning is realized.

The invention provides a video interaction inquiry method based on mass data, and is applied to the field of urban safety protection. The method comprises the steps of constructing a video object space positioning coordinate system taking a single monitoring camera as a center; constructing a video data relation among multiple monitoring cameras; adding a space coordinate in the video data; performing real-time compression on the video data; extracting video semantic features in a compression process, and generating a semantic index; performing storage according to a uniform coordinate system and a uniform video format, wherein the video format at least comprises the photographing time, the space coordinate and the semantic index; performing inquiry by taking an input semance, the photographing time or/and the space coordinate as keywords; searching the video data related to the keywords, and outputting an inquiry result. The invention also provides a corresponding inquiry system. Due to precise space positioning, property-space based bidirectional interaction inquiry for the mass video data is realized.

The invention provides a method for generating a video abstract of a video signal, comprising a video decoding step, a subtitle characteristic acquiring step and a video abstract generating step, wherein the video decoding step is used for decoding the video signal in order to obtain a multi-frame video decoding image; the subtitle characteristic acquiring step is used for acquiring at least one of characteristics of every subtitle in all the subtitles included in the multi-frame video decoding image, wherein the characteristics are as follows: the lasting time of the subtitle, the position of the subtitle in the corresponding video decoding image, the character size of the subtitle, recognition codes and reliability of the subtitle image after optical character recognition (OCR); and the video abstract generating step is used for processing at least one characteristic of the acquired subtitle according to the relationship between the content of the video signal and the characteristics of the subtitle so as to generate the video abstract adaptable to the content of the video signal. By using the method of the invention, the video abstract capable of correctly reflecting the main contents of original video files is generated. The invention further provides a device for generating the video abstract of the video signal and an image processing system with the device.

The invention relates to an electric power communication network fault analysis and positioning method based on a convolutional neural network. The method is mainly used for feature diagnosis of alarmdata and a network topology structure in an electric power communication network. Keyword selection and standardization are carried out on alarm information, synchronous processing is carried out toobtain alarm transaction data, and weighted encoding is carried out on the alarm transaction data and a network adjacent matrix to obtain a fault state matrix containing a network topology connectionrelationship and a network fault state. According to the method, the network topology connection relationship and the alarm information are mapped into the fault state matrix, so that multiple characteristics required by fault diagnosis are improved, and meanwhile, the network topology information is combined, so that the relevance and change adaptability of the network are enhanced.

The invention relates to the technical field of remote monitoring of soil moisture status, in particular to a method for monitoring soil moisture status in whole growth period of large-area crops on the basis of the modified NDVI (normalized difference vegetation index) time sequence. Firstly, region division of farmland, division of the crop growth period and source data acquisition and processing are performed; with the NDVI as a basic soil moisture status index, corresponding factors are introduced for different regions and at different growth periods, and self-adaptive modification is performed on the NDVI; a modified NDVI time sequence growth curve of the crops in the whole growth period is established on the basis of correlation verification of the modified NDVI and actually measured soil moisture status; anomaly analysis is performed on the modified NDVI curve of a monitored year and the average value curve of ordinary years, space-time distribution conditions of the relative soil moisture status, compared with the ordinary years, of the farmland in the monitored year are obtained, the soil moisture status of the farmland and the space-time distribution of the soil moisture status are effectively monitored, and the technical problem of failure in monitoring and back analysis of soil moisture status in whole growth period of large-area crops due to the fact that single NDVI cannot accurately reflect the growth states and the root soil moisture status of the crops in the whole growth period is solved.

The invention relates to the technical field of mobile computing, and provides a hot news sharing method based on geographical information, an apparatus and a system. The method comprises the following steps: the geographical information of users is obtained; positioning is carried out on the users; news events around the users are searched and pushed to the users; if news published by the users is received, the news is bound to the position of a publisher and is stored; and steps such as positioning are executed after timed updating of the geographical information of the users. In the technical scheme of the invention, through real time computing of the geographical information of the users, news publishing sites and the geographical positions of the users and/or focus of attention of the users are associated, and the hot news occurring around the users are shared to the users in real time. According to the technical scheme of the present invention, aiming at a characteristic that the mobile terminal users frequently change positions, events with high correlation degrees are timely and accurately pushed to the mobile terminal users, so that the mobile terminal users can still obtain the news in a rapid and highly efficient mode in cases that screens are relatively small, and the network access speed and flow are limited.

The invention discloses a picture switching method and a picture switching device for picture video playing in an embodiment. The method comprises: the picture switching device obtains the background music of a target picture video and played pictures contained in the target picture video; the characteristic moments contained in the background music are determined according to the characteristic information of the background music; (N-1) characteristic moments are selected from the characteristic moments contained in the background music as switching moments for switching the played pictures in the target picture video; and the played pictures are switched according to the determined switching moments when the target picture video is played. According to the picture switching method, the picture switching device is capable of determining the switching time according to the characteristic moments of the background music, and therefore, the correlation of the background music and picture switching is enhanced, and furthermore, the picture switching modes are not single any more when the picture video is played; hence, the experience of a user is enhanced.

The invention discloses a speech intelligibility measuring method based on an auditory masking effect. The speech intelligibility measuring method based on the auditory masking effect is characterized by comprising the steps that (1) the auditory masking threshold of reference signals and the auditory masking threshold of output signals are estimated respectively according to a psychological acoustic model in the ISO/IEC11172-3 standard in the mode of carrying out frame division on the reference signals and the output signals obtained by enabling the reference signals to pass through a communication transmission system or a speech transmission system, and using the frame as a basic unit; (2) audible components in the reference signals and audible components in the output signals are extracted according to the auditory masking thresholds; (3) the amplitude of the audible components of the reference signals and the amplitude of audible components of the output signals are converted into the loudness according to the ISO226-2003 standard equal-loudness level curve, the sensing similarity, the speech information fidelity and the sensing signal to noise ratio of signals in each frame are calculated, and then the degree of distortion of the output signals is obtained; (4) intelligibility prediction is carried out on the output signals according to a curve of the relation between the degree of distortion and intelligibility. Compared with a common speech intelligibility objective measuring method at present, the speech intelligibility measuring method based on the auditory masking effect is higher in measuring accuracy.

The invention provides a method and a device used for providing search results on a mobile terminal. The device for providing results acquires a plurality of candidate pages and corresponding terminal support information, and page terminal support information is built or updated according to the candidate pages and the corresponding terminal support information. The method further includes conducting follow-up processing on the search results according to terminal attribute information of the mobile terminal that a user corresponds to and the page terminal support information and providing the search result to the corresponding application. Compared with the prior art, the method and the device conduct follow-up processing on the search results of the user by building or updating the page terminal support information for the candidate pages, thereby enabling the search results of the user to meet user need, improving relevance of the search results and a user device, improving information acquisition efficiency of the user and resource utilization ratio of the mobile terminal and improving use experience of the user.

The present application proposes a document topic mining method and apparatus. The method comprises: according to a preset topic mining number, performing loop iteration processing on information in at least one received document based on a probabilistic latent semantic analysis model, and acquiring a posteriori estimate of each topic implied by each sentence in each document; according to the posteriori estimate of each topic, acquiring a membership weight of each word in each topic in each sentence; and generating a topic set corresponding to the topic mining number, wherein each topic set comprises a word related to each topic and screened out according to the membership weight of each word in each topic in the sentence. According to the document topic mining method and apparatus provided by the present application, the document topic is more comprehensively and accurately mined based on a PLSA (Probabilistic Latent Semantic Analysis) algorithm, and the correlation of document topic content is improved, thereby enabling a result of a search engine to be closer to semantic information of the document.

Navigate pressure sensor of internal-combustion engine uses electromagnetic element to measure its admission pressure. Set a diaphragm 5 between main body 1 and end cap 12. Inner space of 1 is divided into a negative pressure chamber B1 and a stable pressure chamber C1 by a spring base 2, which has a damped pore 17. B2 connects with blow hole A1. A spring with gather elasticity is set between 2 and 5. The other side of 5 has a magnet 15, to which sets a Hall element 14 in 12. The invention has merits of sensitive reaction, high action precision, extensive measuring range, high working stability, small capacity, low cost and convenient equipment. It can reflect workload change of internal-combustion engine exactly and can be applied to determine admission pressure and workload of various internal-combustion engines.

The invention discloses a security verification method and system for a mobile phone bank customer. The method includes the following steps that: equipment hardware information and customer number of the customer are obtained; the binding relation between the equipment hardware information of the customer and customer registration information is verified according to the equipment hardware information and the customer number; and a static login password inputted by the customer is obtained, and the static login password is verified; after the binding relation and the static login password both pass verification, it is determined that the login of the customer is successful. With the security verification method and system for the mobile phone bank customer of the invention adopted, the relevance of the customer and equipment is enhanced through the binding relation between the customer information and the equipment hardware information, and harm caused by the stealing of the account password of the customer by criminals can be decreased.

The invention discloses an intra-frame chroma prediction mode selection method, an image processing device and a storage device, and the method comprises the steps: constructing a candidate list of chroma prediction modes, the candidate list comprising a plurality of chroma prediction modes; selecting a part of brightness prediction mode of the brightness block corresponding to the current chromablock as a replacement mode; replacing at least one chroma prediction mode in the candidate list with the replacement mode; and selecting an optimal prediction mode from the replaced candidate list topredict the chroma of the current chroma block by using the optimal prediction mode. In this way, the coding compression rate can be increased.

An embodiment of the invention discloses an intelligent music recommendation system. The system comprises an initialization module, a playing module and an adjustment module, wherein the initialization module is used for constructing a song distance network and initializing a personal song library; the playing module is used for judging a current scene, obtaining a song playing probability according to a weight value, corresponding to the scene, of a song in the personal song library, playing the song, obtaining user feedback and modifying the weight value; and the adjustment module is used for adjusting the song distance network and the personal song library. According to the embodiment of the invention, a music network with a numerical associative relationship among music is established, songs highly associated with a seed song are found in the network, and the personal song library is established by the songs; the personal song library can be intelligently adjusted, so that the song library can be increasingly closer to user preferences, the accuracy of music association values can be improved, and a personalized song library is generated for users; and the recommended songs can be automatically adjusted according to the user preferences, so that the correlation degree of recommendation is increased.

The invention relates to a table structure recognition method and device, a storage medium and electronic equipment, a table image corresponding to a to-be-recognized table can serve as input of a target detection model, the position feature and the element category of each table element in the to-be-recognized table are obtained, and the element category comprises text lines; character recognition is performed on the character lines through a character recognition model to obtain a text semantic feature of each character line; feature extraction is performed on the table image through the image feature extraction model to obtain an image feature map, and the image feature map is samplied to obtain a target image feature corresponding to each table element; according to the position features, the text semantic features and the target image features, target relation features used for representing the topological relation between every two table elements in the to-be-recognized table are obtained through a preset relation extraction model, a graph adjacency matrix used for representing the table structure of the to-be-identified table is determined by a preset classification model according to the target relation characteristics.

The embodiment of the invention provides an interaction content display method and device. The method comprises the following steps that: a client side responds to an interaction operation triggered by a user, and determines the operation identification of the interaction operation; the client side also can obtain the first user level, which corresponds to a live website, of the user, and determines interaction contents corresponding to the interaction operation according to the operation identification and the first user level; and finally, the client side displays the interaction contents ina live picture in a preset display form corresponding to the first user level. By use of the above interaction content display method, substantially, an interaction result generated by the interaction operation is that the interaction contents and the display form of the interaction contents are simultaneously associated with the user level of the user. By use of the above method, a watching userand a host in a live room can be guaranteed to visually know the user who generates the interaction operation according to the interaction contents and the display form of the interaction contents, and association between the interaction result generated by the interaction operation and the user who triggers the interaction operation can be improved.

The present invention provides a robot route planning method by employing an improved convolutional neural network based on a K mean value. The method comprises the following three steps: the step 1, establishing a K mean value clustering algorithm model and a BP network; the step 2, inputting the initial patrol point position of the robot into the K mean value clustering algorithm model; and the step 3, taking the output result of the K mean value clustering model as the input value of the BP network, and performing training, wherein the output value of the BP network is namely the optimal robot patrol route. According to the method provided by the invention, the first layer of the network employs K mean value unsupervised learning, and the second layer of the network employs the BP algorithm supervised learning to increase the correlation of the features obtained through the training and the actual tasks, improve the accuracy, reduce the parameters required training and solve the gradient dispersion problem. The improved algorithm is applied to the robot route plan and has high accuracy and faster training operation speed compared to the main algorithms about the experiment results.

The invention provides a method and device for determining a second query term related to a first query term based on natural searching results. The method comprises the first step of obtaining the natural searching results corresponding to the first query term inputted by a user and the natural searching results corresponding to one or more candidate query terms respectively, the second step of determining the relevancy between the first query term and the one or more candidate query terms respectively according to check processing of the natural searching results corresponding to the first query term and the one or more candidate query terms, and the third step of determining the second query term related to the first query term from the one or more candidate query terms according to the relevancy. Compared with the prior art, the relevancy between the second query term and the first query term inputted by the user is guaranteed, furthermore, the relevancy between the displaying results corresponding to the second query term and the first query term is improved, and therefore the usage experience of the searching user is improved.

The invention provides a word vector webpage intrusion detection method based on semantic analysis, which relates to the field of webpage attack detection, and comprises the following steps: 1) crawling webpage source codes corresponding to URLs of websites, screening and classifying the webpage source codes and taking the webpage source codes as a training set; 2) extracting Chinese character strings of all webpage source codes in the training set to perform word segmentation processing; 3) respectively carrying out keyword screening on word segmentation processing results and constructing akeyword library; 4) constructing a word vector of the keyword, and performing normalization processing; 5) calculating the weight in combination with the normalized TF-IDF to establish a weight word vector space, 6) training a webpage intrusion detection model, 7) judging whether the webpage to be detected is invaded or not through the model. The invention provides a webpage intrusion detection method for judging whether a webpage is maliciously invaded by vocabularies by utilizing a model.

The invention discloses a JPEG coding method lowering image diamond effect under a low compression code rate. A full-phase anti-discrete cosine bi-quadrature transformation replaces discrete cosine transform algorithm in conventional JPEG coding, and uniform quantification is employed, thereby solving the problem of severe loss of high frequency coefficient generated during increasing a data compression ratio in conventional JPEG coding, better recovering image details corresponding to the high frequency coefficient during reconstruction of images, and solving the problem of obvious diamond effect under low compression code rate. The coding method ensures coding image quality and overcome the defects of low JPEG coding compression ratio in the prior art, so that image documents are more suitable for transmission with narrow bandwidth. The JPEG coding method has promising prospect in an image processing field of a wireless multimedia sensor network.

The invention provides an associated problem recommending method, device and system.The method comprises the steps that segmentation is performed on an input problem of a user, and a segmentation result is obtained; according to the segmentation result, a headword and other keywords of the input problem are determined; according to the headword and at least one associated rule of the input problem, an associated problem associated with the input problem is retrieved from a problem base; the associated problem is recommended to the user.According to the associated problem recommending method, device and system, the associated rules are obtained through an associated rule algorithm, and the problem, not limited to synonymous problems or problems obtained through the preset rules, associated with the input problem of the user is analyzed by means of the associated rules, so that the relevance of the associated problems is improved, the recommended result is more accurate and comprehensive, and the user experience is improved.

The invention discloses an accelerated test method for simulating atmospheric corrosion. The method specifically comprises the steps: preparing a spray solution through NaSO4 and NaCl, carrying out asalt spray test on a simulation part through the spray solution, and then performing a drying test on the simulation part; performing a humidity test on the simulation part, and finally performing a salt spray test, the drying test and the humidity test repeatedly and alternately. According to the accelerated test method for simulating atmospheric corrosion provided by the embodiment of the invention, Na2SO4 and NaCl are used for preparing the spraying solution; the pH value can be adjusted according to the environment of the simulated region, and the erosion effect of SO42- and Cl- on the simulation part can be simulated according to the contents of SO42- and Cl- in the simulation region environment; and the salt spray test, the drying test and the humidity test are alternately carried out to simulate the erosion effect of different climates or weathers in the natural environment on the simulation part, thereby facilitating the improvement of the correlation between the acceleration test and the erosion in the natural environment.