103 results about "Document representation" patented technology

A method, system and computer program product for generating a document representation are disclosed. The system includes a server and a client computer, and the method involves: receiving into memory a resource containing at least one sentence of text; producing a tree comprising tree elements indicating parts-of-speech and grammatical relations between the tree elements; producing semantic structures each having three tree elements to represent a simple clause (subject-predicate-object); and storing a semantic network of semantic structures and connections therebetween. The semantic network may be created from a user provided root concept. Output representations include concept maps, facts listings, text summaries, tag clouds, indices; and an annotated text. The system interactively modifies semantic networks in response to user feedback, and produces personal semantic networks and document use histories.

The present invention facilitates the specification and distribution of templated content materials by a content provider over an information exchange network such as the Internet. The present invention incorporates a system for managing inventories of graphical elements and their relationships to pre-defined page templates. A database capable of keeping track of users and their corresponding access privileges within the system is employed to monitor user activity. Ultimately, through the use of a software component delivered over the Internet for use within standard web browsers, end-users are able to populate templates under the constraints imposed by the rules of the manufacturers at the time of template design. These population elements which “fill in the blanks” of the pre-defined templates may be either of type IMAGE or TEXT. Image regions are populated by choosing from a subset of the entire image inventory, while TEXT types can be completely free form, with specific rules guiding justification, point size, font, and leading, or “fill in the blank” form with the same constraint rules as free form. Once the end user has met all of the criteria for a fully populated template, the system provides sophisticated means for downloading a high resolution file (such as a print-ready file or other file representation of the composed publication) which encapsulates all resources needed (layout, images, fonts, and constraint geometries) to fulfill the requirements of the publication. The downloaded file may be printed or published by electronic transfer, e.g., to a publisher for printing of the actual publication.

Provided are a computer program product, system, and method for using a metadata image of a file system and archive instance to restore files in the file system. A metadata image of the file system for a point-in-time backup as of a point-in-time includes information on files and directories in the file system as of the point-in-time and an archive instance including a copy of database records in the backup database for the files in the point-in-time backup. A restore request is received. A file representation is created of each file to restore in the directory structure of the file system from the metadata image, wherein at least one of the created file representations indicates that the file is stored off-line and has an external identifier used to access information on the file in the database records in the archive instance for the point-in-time backup.

An exemplary system includes a browser to browse a web page based on a web page definition having a slicing tree defining an arrangement of rectangular regions in the web page. The web page definition can include parametric data describing adaptability parameters associated with a rectangular region. A rendering module renders an adapted web page based on the web page definition, and a proxy module generates an intermediary adapted web page definition. A method includes rendering the web page according to a slicing tree and block property data in an associated web page definition. The method may include determining a set of unsummarized blocks that maximize information fidelity.

A method, system and computer program product for generating a document representation are disclosed. The system includes a server and a client computer, and the method involves: receiving into memory a resource containing at least one sentence of text; producing a tree comprising tree elements indicating parts-of-speech and grammatical relations between the tree elements; producing semantic structures each having three tree elements to represent a simple clause (subject-predicate-object); and storing a semantic network of semantic structures and connections therebetween. The semantic network may be created from a user provided root concept. Output representations include concept maps, facts listings, text summaries, tag clouds, indices; and an annotated text. The system interactively modifies semantic networks in response to user feedback, and produces personal semantic networks and document use histories.

Methods and systems for managing open documents are disclosed. Document representations are displayed in a carousel display. Each of the representations displays a document viewport portion of content from a corresponding open document. Upon determining a first gesture associated with a selected representation, a full view of the document viewport portion of the open document corresponding to the selected representation is displayed. The content of the open document displayed in the document viewport portion may be adjusted based upon a user action in the open document. Upon determining a second gesture, the full view of the document viewport portion is closed and the adjusted content is displayed as the document viewport portion in the carousel display. A greater portion of the open document than what is visible in the document viewport portion is displayed.

The life history of a person or entity can be presented in a graphical representation of a highway. Life events may be represented by simple data strings, or by files such as photographs, dissertations, job offers, and love-letters, among others. For ease in viewing, the information representing the life history is categorized according to type (medical, educational, photographic, etc.) and placed in lanes corresponding to the type of information. The information is also organized by date, being placed between mile corresponding to temporal periods, for instance, years. Other graphical arrangements of stored information are also included.

A computer method, system and code, for representing a natural-language document in a vector form suitable for text manipulation operations are disclosed. The method involves determining (a) for each of a plurality of terms selected from one of (i) non-generic words in the document, (ii) proximately arranged word groups in the document, and (iii) a combination of (i) and (ii), a selectivity value of the term related to the frequency of occurrence of that term in a library of texts in one field, relative to the frequency of occurrence of the same term in one or more other libraries of texts in one or more other fields, respectively. The document is represented as a vector of terms, where the coefficient assigned to each term includes a function of the selectivity value determined for that term.

Document retrieval system and method are disclosed which can diminish a gap between the user's retrieval intention in information retrieval and the configuration of a query as well as document representations in database and which permits easy retrieval reflecting the user's retrieval intention. The user enumerates a group of words which the user hits upon, as a primary query. Upon receipt of the primary query, the system estimates relational representations which the words (group) of the primary query can possess, and then makes expansion of the query through a partial coincidence of the relational representations and sample spaces extracted from document data to prepare a query candidate representation group. The expanded query candidate representation group is presented to the user. The user then simply chooses a relational representation candidate in accordance with his or her intention. A retrieval execution query is constituted by the thus-selected representation.

The invention belongs to the technical field of cross-media correlation learning, and particularly discloses a deep cross-mode correlation learning-based image retrieval method for a free-hand sketch.The method comprises three main algorithms of deep multi-mode feature generation, multi-mode correlation learning modeling and similarity sorting optimization. By utilizing a deep learning technology, a depth semantic feature and a depth visual feature are constructed for describing a text tagging part and an image / sketch part in a multi-mode document. Based on a multi-mode document representation, a cross-mode correlation model is built for modeling a whole multi-mode document set, thereby describing correlation among different modes of the multi-mode document. Based on correlation featuresobtained after correlation modeling, retrieval results are sorted and optimized, and color images and texts with the maximum similarity with the queried sketch are returned.

The invention provides a document representation method and a document analysis method including extraction of important sentences from a given document and / or determination of similarity between two documents.The inventive method detects terms that occur in the input document, segments the input document into document segments, each segment being an appropriately sized chunk and generates document segment vectors, each vector including as its element values according to occurrence frequencies of the terms occurring in the document segments. The method further calculates eigenvalues and eigenvectors of a square sum matrix in which a rank of the respective document segment vector is represented by R and selects from the eigenvectors a plural (L) of eigenvectors to be used for determining the importance. Then, the method calculates a weighted sum of the squared projections of the respective document segment vectors onto the respective selected eigenvectors and selects document segments having the significant importance based on the calculated weighted sum of the squared projections of the respective document segment vectors.

A method and a system allow presentation of web pages to an internet appliance (e.g., a hand-held computer, a mobile telephone, or a digital personal assistant) according to user preferences. The user preferences are captured by a management server, which provides a web page customization service in conjunction with a document manager, which parses the web pages to identify information units. The customized web pages are stored in a database using a standardized hypertext document representation device, such as XML. The customized web pages are accessible from a portal adapted for accessing by the internet appliance. In one implementation, the user is also offered pre-configured resources for frequently used services when accessing the portal using the internet device.

A system to catalog and search point-in-tine instances of a file system is disclosed. A catalog engine takes backups of file data generated by a storage system and catalogs the backups of file data into a searchable catalog of independent metadata records. The metadata is represented by baseline structure and delta files.

The invention discloses a title generation method based on a variational neural network subject model, belonging to the technical field of natural language processing. This method automatically learnsthe document topic hidden distribution vector by variational self-encoder, and combines the document topic hidden distribution vector and the document representation vector learned by multi-layer neural network with attention mechanism, so as to express the comprehensive and deep semantics of the document on the topic and global level, and to construct a high-quality title generation model. Thismethod uses the multi-layer encoder to learn the more comprehensive information of the document, and improves the effect of summarizing the main idea of the full text of the title generation model; the topic implicit distribution vector of VAE learning is utilized, and the document content is represented in the abstract level of topic. The topic implicit distribution vector and the document information learned by the multi-layer encoder are combined with the deep semantic representation and context information to construct a high quality title generation model by using the attention mechanism.

A machine-oriented notation for representation and interchange of extensible documents: a method, system, and computer program product for operating upon (e.g. parsing, and storing documents in) this notation. The notation, referred to herein as “mXML” for “machine-oriented XML”, is designed to be more compact than the Extensible Markup Language (XML), while still conveying the content and semantics of the data and the structure of the document. Documents may be created directly in mXML. In general case, a document represented using mXML notation can be processed more efficiently than when using the existing human-friendly XML notation, requires less storage space, and has a lower transmission cost for data interchange. XML documents can be converted to mXML using techniques of the present invention, and vice versa. Techniques disclosed herein are also applicable to notations other than XML.

Methods and apparatus are described for classifying documents using a document representation model based on implicit user feedback obtained from search engine queries. The model may be used to achieve better results in non-supervised tasks such as clustering and labeling through the incorporation of usage data obtained from the search engine queries.

An apparatus for generating a collection profile of a collection of different media data items has a feature extractor for extracting at least two different features describing a content of a media data item for a plurality of media data items of the collection, and a profile creator for creating the collection profile by combining the extracted features or weighted extracted features so that the collection profile represents a quantitative fingerprint of a content of the media data collection. This collection profile or music DNA can be used for transmitting information, which is based on this collection profile, to the entity itself or to a remote entity.

The invention provides a text classification method based on deep multi-task leaning. The method comprises the steps that by means of a recurrent neural network obtained through other task training and by combining the learning ability of a convolutional neural network, additional document representation is obtained, that is to say, a large amount of external information is introduced, semantic representation of a document is extended, and the problem that training data is insufficient is effectively solved. Accordingly, compared with a traditional multi-task leaning method, the convolutional neural network is used for conducting feature extraction on bottom-layer features of an auxiliary task, the features of other tasks can be utilized for being effectively transferred to the current task, and the performance of text classification is improved.

Methods and systems for managing open documents are disclosed. Document representations are displayed in a carousel display. Each of the representations displays a document viewport portion of content from a corresponding open document. Upon determining a first gesture associated with a selected representation, a full view of the document viewport portion of the open document corresponding to the selected representation is displayed. The content of the open document displayed in the document viewport portion may be adjusted based upon a user action in the open document. Upon determining a second gesture, the full view of the document viewport portion is closed and the adjusted content is displayed as the document viewport portion in the carousel display. A greater portion of the open document than what is visible in the document viewport portion is displayed.

A method for determining, from a document, an answer to a query using a query answering system, comprising: (i) encoding, using an encoder, one or more documents; (ii) encoding a received query; (iii) generating, using an attention mechanism, a query-aware document representation comprising alignment between one or more words in one of the plurality of documents and one or more words in the query; (iv) generating, using a hierarchical self-attention mechanism, a word-to-sentence alignment of the query-aware document representation; (v) labeling, using a conditional random field classifier, each of a plurality of words in the word-to-sentence alignment with one of a one of a plurality of different sequence identifiers, resulting in possible labeled answering spans; and (vi) generating, from the one or more possible labeled answering spans, a response to the query.

The life history of a person or entity can be presented in a graphical representation of a highway. Life events may be represented by simple data strings, or by files such as photographs, dissertations, job offers, and love-letters, among others. For ease in viewing, the information representing the life history is categorized according to type (medical, educational, photographic, etc.) and placed in lanes corresponding to the type of information. The information is also organized by date, being placed between mile corresponding to temporal periods, for instance, years. Other graphical arrangements of stored information are also included.

The invention discloses a text document representation method and a device based on depth learning topic information enhancement. The method comprises the following steps: S1, data preprocessing operation is carried out on the corpus document in the form of text. S2, a text sequence layer is designed, and the context information of each word in the word order is embedded into the representation vector of each word in the document. S3, the sequence elements are transitioned to higher-level topic information through the attention layer. S4, in the topic layer, a representation of the current document D in all topic directions is generated. S5, the similarity between all the topic information is limited. S6, the topic representation vector is fused into the semantic representation vector Repof the document D at the presentation layer. 7, that parameters of the Rep are updated by a classify and an objective function, the method can efficiently embed the context semantic information and the potential topic information of a text sequence into a document representation vector, and the presentation vectors enhanced by the topic information can significantly improve the performance of a text mining model use the Rep.

The invention provides an entity question answering method and device based on a neural network and a terminal, wherein the method comprises converting a word contained in a question and a candidate document into a word vector, and generating a corresponding question word vector sequence and a candidate document word vector sequence; inputting the problem word vector sequence and the candidate document word vector sequence into the long-short-term memory network model, and outputting the word coding sequence of the problem and the candidate document; matching the word encoding sequence of theproblem and the word encoding sequence of the candidate document to generate a candidate document representation based on matching information, wherein the candidate document representation comprisesa plurality of word representations; selecting start and end words from all word representations and generating entity answers based on the start and end words. The method reduces explicit computationand cumulative error, effectively utilizes semantic representation between questions and documents, and improves positioning accuracy of entity answers.

The invention discloses a text representation method and device based on a hierarchical neural network. The method comprises: converting each word forming a sentence into a vector; Inputting vectors corresponding to all words in the sentence into a neural network for aggregation, and outputting sentence representation corresponding to the sentence; Inputting all the sentence representations into aneural network to be aggregated, and generating document representations corresponding to all the sentence representations; And converting the document representation into a document classification vector through a full connection network, and obtaining prediction probability distribution of document classification based on the document classification vector. According to the method and a device,A hierarchical mechanism is introduced into a neural network model to solve a document representation problem for text classification; Interoperability of different tasks is better improved, a hierarchical neural system structure is fused into a neural network method, a new neural network model based on layering is caused, accuracy, performance and the like are obviously superior to those of an existing neural network model, and consumption is lower.

A system and method for creating, managing, and displaying an interactive display for 3D digital collectibles comprising a virtual, three dimensional, n-sided structure including a digital media file or set of digital media files representing an event rendered on a representation of a first surface thereof, and data relating to the event rendered on at least a second surface thereof, where the digital media file may be a video clip of the event that can be played automatically via a media player associated with the display. The interactive display may provide a graphical user interface that displays a set of user tools to interact with the 3D digital collectibles and a user interface control module that receives user input via the user tools and generates instructions to control the display of one or more 3D digital collectible display elements.

An extractive news digest generation apparatus based on an attention encoder includes a statement encoder for acquiring a document and dividing the document into a plurality of sentences; a document encoder for acquiring a document representation according to the relationship between the plurality of sentences and the plurality of sentences; a statement extractor for extracting sentences used as summaries from the plurality of sentences and the document representations. The extractive news digest generation apparatus can capture the relationship and dependency between sentences better, so as to extract abstracts accurately and show more abundant information when generating abstracts.

The invention relates to a cross-domain emotion classification system based on a hierarchical attention mechanism, and the system comprises a text preprocessing module which is used for the characterization of a cross-domain text; a pivot feature extraction module, used for learning a feature representation space adapted to the field to obtain pivot feature document representation of the source field and the target field; a non-pivot feature extraction module, used for acquiring non-pivot feature representation; and an emotion category output module, used for obtaining a final emotion classification result. According to the method, efficient cross-domain emotion classification is realized, the cross-domain emotion classification precision is improved, and the consumption of manual time andenergy is reduced.

The invention relates to a cross-domain emotion classification system based on attention mechanism fusion. The system comprises a comment text preprocessing module used for obtaining vector forms of texts in a source domain and a target domain; a text semantic learning module which is used for learning a semantic dependency relationship between words; an attention mechanism fusion module which isused for fusing different attention modes to obtain comprehensive weights of words for text classification; a hierarchical attention module which is used for calculating attention weights of the textfrom a word level and a sentence level respectively and judging weights of words for sentence representation and sentences for document representation; and an emotion category output module which is used for obtaining a final emotion classification result by utilizing the classification function. According to the method, the potential universal features of the target domain and the source domain can be automatically extracted, the features are abstracted and combined, and finally the emotion category of the text of the target domain is recognized.

The invention relates to a judicial document paragraph classification method and device, computer equipment and a storage medium. The method comprises the steps of obtaining judicial documents; performing character segmentation on the judicial document to obtain a character matrix; carrying out vector extraction according to the character matrix to obtain sentence representation vectors; splicingthe sentence representation vectors to obtain a document representation vector; inputting the document representation vectors into a classification model for classification to obtain paragraph categories; feeding back the paragraph category to the terminal for the terminal to perform information extraction, wherein the classification model is obtained by training a model composed of a bidirectional recurrent neural network and a conditional random field by taking a document representation vector with a category label as sample data. According to the method, the sentence representation vectorsare classified through the classification model composed of the trained bidirectional recurrent neural network and the conditional random field to obtain the paragraph categories, judicial document paragraphs are automatically classified, the generalization ability is achieved, and the extraction accuracy and recall rate are high.