172 results about "Pagerank algorithm" patented technology

A method for generating an emotion abstract of a dialogue text includes the steps of constructing feature vectors, comprising word characteristics and emotion characteristics, of various sentences in the dialogue text; calculating similarity of the vectors by the adoption of a revised cosine similarity algorithm and then calculating the transition probability of the sentences according to the similarity; calculating weight values of the sentences through a PageRank algorithm, carrying out sorting according to the weight values and selecting the sentences with large weight values and of the number according with a preset number to form the emotion abstract of the dialogue text at last. According to the method, the abstract does not need to be generated manually, manpower and material resources are saved, and meanwhile precision for generation of the emotion abstract of the dialogue text is improved.

The invention discloses a personalized item-level vertical pagerank algorithm iRank, and relates to a personalized item-level vertical search engine recommendation algorithm iRank. The basic idea of the algorithm is that: deep and detailed researches are made on the vertical field item-level pagerank algorithm and user behaviors, an industry ontology base model is constructed by collecting the information of a vertical field, user behavior information is collected by using a sentiment analysis technology and an Eyetrack technology for recording the eyetrack time of an Internet user in a current page in the field of data mining, a user interest model and a similar user model are constructed statistically, and finally the personalized recommendation algorithm based on the user interest model and the similar user model is combined in pagerank. The algorithm iRank can intelligently perform personalized ranking on item-level search engine semantic information retrieval results to influence rank ratings of returning items and realize inter-user personalized ranking on an item set result. Compared with the conventional pagerank algorithm (PageRank, HITS), the algorithm iRank improves the capability of a search engine in coping with user interests, and has a high practical application value.

The invention discloses a method for identifying microblog key users based on an improved Page Rank. The method comprises the steps that microblog information data are input, wherein the microblog information data comprise n microblogs; word segmentation is conducted on texts of the n microblogs; according to a word segmentation result, a reverse index structure is established, so that retrieval is conveniently conducted according to appointed keywords; according to the retrieved relevant microblog, forwarding hierarchy information of the microblog is extracted and a weighting directed graph is established, wherein the weighting directed graph is a forwarding network G; the forwarding network G is divided into a plurality of maximum connected subgraphs Gi; the Page rank algorithm is applied to each sub network Gi according to the parallelization computing technology; computing results of the sub networks are combined, so that ranking results of the whole network G are generated; the first m ranking results of the ranking results are selected, serve as the key users and are output. According to the method for identifying the microblog key users based on the improved Page Rank, the parallelization computing technology is adopted, a dynamic forwarding network of a microblog platform is ranked and computed in a big data environment, so that the key users in the information transmission process are identified, and the method is applied to the fields of network public opinion analysis and the like.

The invention belonging to the field of social network information analysis makes a request of protecting a method and system for predicting information popularity of a social network. The method comprises steps of data acquisition, attribute extraction, model construction and prediction analysis. An information dissemination network is refined by combining user relationships and node behavior data in a social network; attributes affecting the measure of propagation power are extracted by starting with an individual behavior dimension and a node interaction dimension and related definitions are provided; a dual-weighted social network is constructed again and a node propagating power in the network is measured based on an improved PageRank algorithm; and individual characteristics of an information publisher and forwarding characteristics of the information published within one hour are extracted by using the information as a center and training is carried out by using an LR classifierto obtain an information popularity prediction model. Therefore, the information popularity can be predicted effectively; and a network group event is found out timely and important propagation nodesin the information dissemination network are identified.

The invention provides a question and answer community expert recommendation method for performing interest modeling by use of tag fusion across platforms. According to the method, cross-platform common users are utilized to construct word vectors of tags by combining an LDA topic model and word2vec, a tag semantic similarity matrix is constructed for text data of different platforms, a fusion feature space is generated, and a fusion space model of the users is obtained. Compared with a single-network user model, the cross-platform user model can cover different features of the users more comprehensively, and the user features are described more clearly. Meanwhile, answer abilities of the users and cross-platform community influences of the users are comprehensively considered, a PageRank algorithm based on a fusion network is used to perform authority evaluation on the users, and then community feedback is considered to perform ability evaluation on the users. Through experiment comparison with a reference interest model, the single-network user model, a collaborative filtering recommendation model and other algorithms, it is shown that the algorithm has a better recommendation effect.

The invention discloses a PageRank method based on quick similarity. A PageRank algorithm is combined with an improved Hamming distance similarity algorithm, the invention provides a novel webpage sequencing method. The PageRank algorithm is an algorithm for purely researching a webpage link and is easy to generate a problem of topic shift. Aiming at the problem, the algorithm provides two improvements: 1, by combining with the Hamming distance similarity algorithm, similarity of an index word and a webpage text is calculated, searching precision rate is improved; and for improving searching recall ratio, the Hamming distance similarity algorithm needs to be improved, the searching of a synonym of the index word is increased, and the searching range is enlarged. According to the two improvements, a calculating formula of the quick similarity PageRank algorithm is obtained, and the searching requirement is met from the two aspects of the recall ratio and the precision rate.

The invention discloses a scholar-influence evaluation method based on an academic heterogeneous network. According to the method, interaction and communication of academic subjects are analyzed in a dissertation-scholar heterogeneous network, and when the dissertation-scholar heterogeneous network is used for evaluating the influences of scholars, reference relations between a PageRank algorithm and dissertations are used for distinguishing the importance of the dissertations. Meanwhile, in consideration of the characteristic that the number of citations that the dissertations receive needs time to be constantly accumulated, according to the published time of different dissertations, time functions are added, the PageRank algorithm is improved, and the influence of the published time on the dissertation importance is intensified. On the basis of the analysis, the method can evaluate the influences of the scholars in the dissertation-scholar heterogeneous network by a random walk algorithm so that the accuracy and the reliability of the scholar-influence evaluation can be improved, and providesa basis for the scholar-influence evaluation, the examination and approval of scientific-research awards and funds, and other tasks.

The present invention discloses a social network-based user keyword extraction method. The method comprises: obtaining a blog post published by a target user within a preset time interval and performing word segmentation processing to obtain a word list of each blog post; inputting the word list corresponding to each blog post into a Word2Vec model for training to obtain a word vector model; extracting keywords corresponding to the blog post based on a keyword extraction algorithm to form candidate keyword sets for the target user, based on the word vector model, calculating a word vector of each keyword in each candidate keyword set, and constructing a semantic similarity graph; and running the Pagerank algorithm on the semantic similarity graph, and scoring the keywords to obtain interest keywords of the user. The present invention further proposes a social network-based user keyword extraction apparatus and a computer readable storage medium. The technical scheme of the present invention solves the technical problem in the prior art that it is difficult to extract the keywords that can effectively represent the interest of the user, according to the blog post of the user.

The present invention relates to a parallelization critical node discovery method for postal delivery data. The method comprises the following steps: step S1: acquiring node activity according to the total number of sending and receiving times of each node in a set time in the postal delivery data, and taking the node activity as the own weight value of the node; step S2: acquiring the weight values of edges of each node pair according to the interaction frequency and shared neighbor number metric indexes of each node pair in the set time in the postal delivery data, and defining a network formed by the postal delivery data as a directed double-weighted network graph; and step S3: adding the own weight values of the nodes and the weight values of the edges of the node pairs on the basis of a PageRank algorithm, and excavating critical nodes in the directed double-weighted network graph in parallel. In contrast to the prior art, the parallelization critical node discovery method fully utilizes information in a logistics postal delivery network, reduces the loss of useful information, improves the accuracy of discovery of critical nodes in the network, and parallel operation is implemented at the same time, thereby greatly improving the efficiency and stability of critical node excavation.