Abstract: A non-transitory computer-readable recording medium stores an index generating program that causes a computer to execute a process including: generating presence information of a plurality of pieces of text data, the presence information including whether each of a plurality of elements, included at least one of the plurality of pieces of text data, is present for each of the plurality of pieces of text data, the presence information including a first axe for the plurality of elements and a second axe for the plurality of pieces of text data; detecting collision data for hashed index information when generating the hashed index information, the collision data corresponding to data elements that are independent in the presence information; and setting additional values to each of a plurality of specific collision data, respectively, for one of the plurality of hashed axes.

Abstract: Systems and methods are disclosed that provide high-level, ontology-based analysis of low-level data stored within an unstructured key/value store. The systems and methods allow an analyst to make sense of massive amounts of data from diverse sources without having any knowledge of the underlying physical data storage. The systems and methods provide flexible ontology assisted addressing, embedding such addressing in existing query languages such as widely used Structured Query Language (SQL), and returning results and provenance information of the results.

Abstract: A technique for suggesting patterns to search documents for information of interest includes acquiring a working set of spans for a document set that includes one or more documents. A list of one or more suggested patterns is generated by applying a pattern suggestion algorithm (PSA) to the set of spans for each document in the document set. One or more unique patterns are generated by applying a pattern consolidation algorithm (PCA) to the generated list of suggested patterns. Pattern information for each of the unique patterns is then generated. The pattern information includes a respective first count that corresponds to the number of times each of the unique patterns occurs in the document set.

Abstract: A computer system uses a search engine to search a dataset using an initial query provided by a user and thereby to produce initial search results. The system enables the user to select portions of the initial search results. The system identifies metadata associated with the selected portions, and displays information representing that metadata in a form that is easily understandable by a person not trained in the annotation system in which the search results are encoded. The user may instruct the system to add one or more of the displayed metadata elements to the initial query, in response to which the system may add the specified metadata elements to the initial query to produce a modified query. The system may search the dataset using the modified query and thereby produce modified search results. The process may be repeated as desired by the user to further refine the search results.

Abstract: A system for performing semantic search receives an electronic text corpus and separates the text corpus into a plurality of sentences. The system parses and converts each sentence into a sentence tree. The system receives a search query and matches the search query with one or more of the sentence trees.

Abstract: An aspect of topic generation includes parsing communications conducted by users through an application. The communications include a communication generated by a sender and response communications received from recipients in reply to the communication generated by the sender. An aspect also includes identifying keywords, timestamps, and indications of sentiment from the parsed communications through natural language processing, determining a focus of the communication generated by the sender based on the keywords identified from the parsing, and formulating a topic for a publication based on criteria including a frequency of occurrence of the keywords in the parsed communications, a threshold level of the indications of sentiment that appear in the parsed communications, and/or a number of the communications containing one or more of the keywords having corresponding timestamps that fall within a threshold period of time. An aspect further includes submitting the topic for publication to the sender.

Abstract: A device may receive a first textual document and a set of second textual documents. The first textual document may identify parameters. The device may process the first textual document, based on a natural language processing algorithm, to identify first values corresponding to the parameters. The device may process the plurality of second textual documents, based on the natural language processing algorithm, to identify second values corresponding to the parameters. Each of the plurality of second textual documents may be associated with one or more respective second values of the second values. The device may determine scores for the plurality of second textual documents. A score may be determined based on comparing one or more of the first values to one or more of the second values associated with a second textual document. The device may provide information identifying the plurality of second textual documents and the scores.

Abstract: Provided are a method, computer program product, and system for improving content. Feedback related to the content is received from a reviewer. The feedback is analyzed with text analytics and classified by on the feedback analysis. A reviewer score is generated and a task is generated for reviewing the feedback wherein the task includes the feedback classification and the reviewer score.

Abstract: A method for classifying a new instance including a text document by using training instances with class including labeled data and zero or more training instances with class including unlabeled data, comprising: estimating a word distribution for each class by using the labeled data and the unlabeled data; estimating a background distribution and a degree of interpolation between the background distribution and the word distribution by using the labeled data and the unlabeled data; calculating two probabilities for that the word generated from the word distribution and the word generated from the background distribution; combining the two probabilities by using the interpolation; combining the resulting probabilities of all words to estimate a document probability for the class that indicates the document is generated from the class; and classifying the new instance as a class for which the document probability is the highest.

Abstract: An operation terminal (120) requires an input of an authentication code when an authority granting request for corresponding electronic information to a tag (100) is received from a user, and transmits the authentication code to a server together with the authority granting request. The server (140) has an authentication code memory unit (142) that stores the authentication code for each tag, and an authorized user management unit (144) that manages the authorized user for each tag. In the case where an authority granting request and an authentication code are received from the operation terminal (120), the server (140) registers the user of the operation terminal (120) with the authorized user management unit (144) as an authorized user of the tag (100) under the conditions that an authorized user of the tag (100) is not registered with the authorized user management unit (144) and the received authentication code matches the contents of the authentication code memory unit (142).

Abstract: Techniques of generating a knowledge graph metadata network (metadata network) for digital asset management (DAM) are described. A DAM logic/module can obtain one or more first metadata assets describing characteristics associated with digital assets (DAs) in the DA collection. The DAM logic/module can also determine second metadata asset(s) and third metadata asset(s) describing characteristics associated with DAs in the DA collection based on the first metadata asset(s). The DAM logic/module can generate at least some of the metadata assets as nodes in a metadata network associated with the DA collection. The DAM logic/module can also determine, for at least two of the metadata assets, a correlation between the at least two metadata assets. The DAM logic/module can generate an edge in the metadata network between the nodes that represent the at least two metadata assets to represent the determined correlation.

Abstract: The present disclosure discloses a display control method and apparatus for multimedia data, and belongs to the field of Internet technologies. The method includes: acquiring a ratio of a total real display count to a total planning display count corresponding to each display area within a first time period, determining the ratio as a data display rate of each display area within the first time period, and controlling, according to the data display rate of each display area within the first time period, multimedia data displayed in each display area. In the present disclosure, a data display rate of each display area within a first time period is acquired, and multimedia data displayed in each display area is controlled according to the data display rate of each display area within the first time period, so that stability of accuracy of a control result is improved, and a control effect is enhanced.

Abstract: Embodiments described herein generally relate to a bulk keyword management application. The bulk keyword management application allows a user to associate a plurality of keywords with a plurality of media items at once. The plurality of keywords may be searchably associated with the plurality of media items. Associating keywords to media items in bulk effectively organizes a gallery of media items in a time efficient manner, making the gallery easy to navigate through a search means.

Abstract: A gallery software application enables a user to browse, view, and interact with various content items, such as still images and videos. The gallery includes a graphical user interface with two modes: stack and grid. In stack mode, images (which represent content items) are divided into groups, and a group is represented by a displayed stack of images. In grid mode, images are displayed in a grid of rows and columns. The displayed stacks and images can be scrolled. When the scrolling reaches the end of a set of stacks or images, the last stack or image is displayed with perspective distortion. In grid mode, a value is displayed for meta-data of a content item that is not currently displayed. If a user indicates the displayed value, the grid is scrolled until the content item whose meta-data matches that value is displayed.

Abstract: The present disclosure discloses a searching method and apparatus. A specific embodiment of the searching method comprises: receiving a search request sent by a client, the search request including a target picture; analyzing the target picture to extract text information and an image characteristic; matching the text information with text information of each preset picture in a preset picture set to determine a first similarity, matching the image characteristic with an image characteristic of the preset picture in response to the first similarity greater than a first preset threshold, and determining whether the preset picture is determined as an identical picture based on matching results; and acquiring associated information of the identical picture, sending the identical picture and the associated information as a search result to the client to be displayed on the client. The present embodiment improves the search efficiency and reduces the labor cost of the search.

Abstract: Composite materials and methods for their preparation are described herein. The composite materials can comprise a polyurethane and an absorptive filler. The polyurethane can be formed from the reaction of at least one isocyanate selected from the group consisting of diisocyanates, polyisocyanates, and combinations thereof, and one or more isocyanate-reactive monomers. The one or more isocyanate-reactive monomers can comprise at least one polyol and a first isocyanate-reactive monomer which includes one or more isocyanate-reactive functional groups and a moiety configured to associate with the absorptive filler.

Abstract: A moment identification system automatically generates a playlist of conversations having a specified moment. A moment can be occurrence of a specific event or a specific characteristic in a conversation, or any event that is of specific interest for an application for which the playlist is being generated. For example, a moment can include laughter, fast-talking, objections, response to questions, a discussion on a particular topic such as budget, behavior of a speaker, intent to buy, etc., in a conversation. The moment identification system analyzes each of the conversations to determine if one or more features of a conversation correspond to a specified moment, and includes those of the conversations in the playlist having one or more features that correspond to the specified moment. The playlist may include a portion of a conversation that has the specified moment rather than the entire conversation.

Abstract: A system and method for the caching of files streaming from at least one source-server towards at least one end-user via the internet; system characterized by at least one integrated circuit (IC) configured to manage and direct files to be cached in at least one storage-element, said ICs and storage elements interconnected by at least one high speed bus. The storage elements comprise different types of non-volatile memory, each characterized by its access speed. A protocol selects files and file fragments for caching. At least one database maintains data on each cached file or file fragment, the data comprising: caching status, data format characteristics, and relative end-user demand. A protocol continuously allocates or reallocates memory for the caching of the files and file fragments, from among the storage elements, as a function of the database data and respective amounts of memory available in the storage elements.

Abstract: Determination of a degree of similarity among and between a set of text notation schema instances. One type of text notation schema instance is the JSON type. In some embodiments, the degree of similarity is expressed as a schema variance value which is determined by individually comparing the schema instances of the set of text notation schema instances to a representative majority schema. Also, determining a quality of a data source associated with the plurality of text notation schema instances based, at least in part, upon the similarity value.

Abstract: Approaches for facilitating queries of a topic-based-source-specific search system are disclosed. In certain implementations, an input relating to a query may be received. Suggested ones of predefined sources and suggested ones of information items of the predefined sources may be determined based on the input. A set of suggestions including a group of suggestions relating to the suggested sources and a group of suggestions relating to the suggested information items may be provided for presentation on a user interface. In some implementations, metadata that indicates first, second, or other attributes relating to the predefined sources or the information items may be stored in association with the predefined sources or the information items. A second input relating to the query may be received responsive to providing the set of suggestions. A presentation of a comparison between the first attribute and the second attribute may be provided based on the second input.

Abstract: Recurrent neural networks (RNNs) can be visualized. For example, a processor can receive vectors indicating values of nodes in a gate of a RNN. The values can result from processing data at the gate during a sequence of time steps. The processor can group the nodes into clusters by applying a clustering method to the values of the nodes. The processor can generate a first graphical element visually indicating how the respective values of the nodes in a cluster changed during the sequence of time steps. The processor can also determine a reference value based on multiple values for multiple nodes in the cluster, and generate a second graphical element visually representing how the respective values of the nodes in the cluster each relate to the reference value. The processor can cause a display to output a graphical user interface having the first graphical element and the second graphical element.

Abstract: An indexing system for documents such as web pages divides a document into elements, such as document object model elements. User attention data from prior interactions with the document are analyzed to determine those elements of a document that satisfy a threshold requirement of user attention. Elements meeting the user attention threshold requirement are added to a set of indexable content for the document. Furthermore, document sections are determined based on attention data and each section is indexed separately. Indexing is per section and based only on the indexable content, thereby enhancing the index relevance, increasing the efficiency of search engines and reducing spamdexing.

Abstract: A method includes receiving a search query and a device type identifier from a user device at a processing system. The method also includes identifying a consideration set of application records based on the search query. Each application record in the consideration set has an initial score associated therewith indicating a degree to which the application record matches the search query. For each application record, the method includes determining a device-specific rating of the software application identified in the application record based on the device type identifier by the processing system, and determining a result score of the application record based on the device-specific rating and the initial score. The method further includes generating device-specific search results based on the consideration set of application records and the result scores thereof by the processing system and transmitting the device-specific search results from the processing system to the user device.

Abstract: A machine is configured to access a search phrase or other communicated phrase and deconstruct the accessed phrase into multiple sub-phrases. The machine performs an analysis of n-grams that occur within the sub-phrases, generates a set of potentially different sub-phrases from the n-grams, and selects which data source among multiple available video sources should be accessed for each generated sub-phrase in the generated set. For example, the machine may allocate each generated sub-phrase to a corresponding data source and cause the corresponding data source to execute a query based on its corresponding sub-phrase. Thus, the machine chooses from which data source to obtain partial search results that correspond to each sub-phrase generated based on the analyzed n-grams in the accessed phrase.

Abstract: An application search system stores information describing device capabilities used by one or more applications. The device capabilities describe interfaces with hardware on the user device. The application search system identifies the application capabilities related to applications in any search results. One method that the application search system identifies application capabilities includes identifying function calls to device interfaces used by an application. The application search system may receive user-selected identification of a desired application capability from a search, from a user-selected entry or from a search term, among others. The application capability in a search may be used to score relevance of applications and increase relevance of applications matching the desired application capability.

Abstract: Embodiments of the disclosure are related to distribution of content in response to search signals in an enterprise environment. Embodiments of the disclosure obtain search signals from a client device associated with a user. In response to obtaining such a search signal, content stored on a resource repository can be located. Such a resource repository can be a private repository associated with an enterprise. In response to located content, a notification can be sent to the client device of the user.

Abstract: A Microblog-based event context acquiring method has: an event-related acquiring information of original microblogs and reposted microblogs thereof contained in a predetermined topic within a statistical time period; calculating an event heat time distribution about the topic corresponding to the information of the original and reposted microblogs thereof related to the event within the statistical time period, the event heat indicating a degree of concern of the original microblogs and the reposted microblogs thereof related to the event in a user group; and selecting, from the statistical time period, a plurality of event context time nodes and corresponding node events according to the event heat time distribution about the topic within the statistical time period so as to constitute an event context of the topic. Event context of the event contained in the predetermined topic by means of the microblogs is acquired quickly and accurately.

Abstract: A search continuation device and its operations include, for example, performing a first electronic search based on first search criteria. The first electronic search generates first search results. The operations can further include electronically storing search data associated with the first electronic search (e.g., in a user account). The operations can further include performing a second electronic search with second search criteria substantially similar to the first search criteria. The second electronic search generates second search results. The operations can further include, retrieving the search data (e.g., from the user account), and providing, based on the search data, one or more indicators for presentation with the second search results. The one or more indicators specify continued search information, such as a degree to which content contained within an item of the first search results was previously accessed or a difference between the first search results and the second search results.

Abstract: A method for presenting search results is provided. The method receives a user input search query; obtains, from a search engine, a first set of search results responsive to the user input search query; efficiently identifies a promoted set of search results for promotion using a set of search promotion rules, each of the set of search promotion rules including a set of terms and one or more document identifiers; revises the first set of search results, based on the identified promoted set, to create a second set of search results; and presents the second set of search results, wherein the second set of search results includes the first set of search results and the promoted set in a promoted position.

Abstract: Methods, systems, and apparatus including computer programs encoded on a computer storage medium, for augmenting search engine index that indexes resources from a collection of resources.

Abstract: Resolving a query received from a first node in a network includes accepting, by a second node in the network, ownership of the query from the first node, receiving, at the second node, an identification of a third node in the network, wherein the identification is received from a user of the second node and the user of the second node believes that a user of the third node has information necessary to resolve at least part of the query, and transferring, by the second node, ownership of the at least part of the query to the third node, wherein the accepting, the receiving, and the transferring dynamically generates a data structure that traces a propagation of the query, and the data structure is accessible to an origin of the query.

Abstract: Internet search engines sometimes provide Internet search results referencing webpages that do not contain all search term elements submitted by a user. The user may then click on such Internet search results where the referenced webpages do not contain an important search term element. The present invention is directed to annotating Internet search results to indicate missing search term elements. This facilitates the user to avoid clicking on an Internet search result where the user's search term elements are not present on the referenced webpage, and thereby prevents wasting the user's time. Furthermore, frequently, search term elements are missing on a webpage referenced by an Internet search result but are found on a descendant webpage thereof. Accordingly, the present invention is further directed to annotating an Internet search result consistent with the presence or absence of search term elements on both a referenced webpage and one or more descendant webpages.

Abstract: A method for quantifying localness of content can be organized into several stages of information acquisition and processing, with each stage focusing on filtering or qualifying content based on geographic information relevant to a place. The method may comprise bootstrapping an initial set of terms for the place, building a local content corpus for the place utilizing the initial set of terms, and populating an index with information from documents in the local content corpus. In response to a request about the place, a query is formed and provided to the index for retrieving local content relevant to the place.

Abstract: A method of stacking tabs in a browser window executable by an electronic device includes, in a first tab associated with a first web resource, receiving a user request for opening a second web resource. A type of the first web resource is identified, the type being one of a hub type and a node type. Responsive to the first web resource being of the node type, the second web resource is opened in the first tab replacing the first web resource. Responsive to the first web resource being of the hub type, a second tab is created, the second web resource is opened therein, and the second tab is stacked on the first tab, the first tab being thereby stacked under the second tab. The first and second tabs are comprised in a tab stack of the browser window, the tab stack comprising a plurality of stacked tabs.

Abstract: Systems and methods are provided methods for providing an online interface for displaying hierarchically or logically arranged user-generated content. One method including receiving, from a first user over a network, topical user-generated electronic content, the topical user-generated electronic content being related to a topic of a web page; displaying the topical user-generated electronic content received from the first user on the web page; receiving, from a second user over the network, responsive user-generated electronic content and an indication of a desire to post the responsive user-generated electronic content on the web page in association with the topical user-generated electronic content; and displaying, adjacent to the topical user-generated electronic content on the web page, a user element or link that a third user can manipulate to display the responsive user-generated electronic content adjacent to the topical user-generated electronic content.

Abstract: A system for automatically generating a wrapper for an entire website, the wrapper characterizing the structure of the website, the system having a plurality of functional elements, including at least one annotation module to classify components of a page and generate an annotated, a page classification module to identify functional and informational components of an annotated page, and an action module to identify an action to be taken to further navigate the website, wherein at least one of the annotation module, page classification module and action module is operable in response to a plurality of domain-specific rules, where a domain is understood as a conceptual domain such as real estate, used cars, or electronics.

Abstract: Systems and methods of operating a web portal environment are disclosed. One such computer-implemented method includes: triggering rendering of a content item being specified by a feed-driven list definition defining a new web content type; reading entries of the content item by a first tag that creates a request attribute containing query relevant information; listening to the request attribute; creating a query; passing the query to a feed integration service; executing the corresponding remote API call; receiving a response from an external data provider; passing the response to the feed-driven objects extension plug-in; creating a rendering context which can be accessed by a second tag; and inserting the external data into markup by the second tag based on a set of predefined feed-driven list presentation components.

Abstract: Representative implementations provide devices and techniques for automatically generating a skeleton of at least a portion of a web service application. The skeleton comprises a framework that may be modified as needed for a desired use, such as for a relational database backed application, for example. In one aspect, the framework includes a versatile model to support different validation methods and varied customization. Filter support allows a user to perform actions on and/or define rules for inputs. In another aspect, the skeleton is automatically activated in an in-memory database repository.

Abstract: A method including: receiving at a server a first request from a first provider to retrieve a first configuration for a first lookup order; determining the first configuration for the first lookup order using the server based at least in part on both an ordering specified by the first lookup order and the first call being through the direct dependency relationship; providing the first configuration to the first provider; receiving at the server a second request from the first provider to retrieve a second configuration for a second lookup order; determining the second configuration for the second lookup order using the server based at least in part on both an ordering specified by the second lookup order and the second call to the first provider by the application being through the transitive dependency relationship; and providing the second configuration to the first provider. Other embodiments are described.

Abstract: Optimization of optical parametric models for structural analysis using optical critical dimension metrology is described. A method includes determining a first optical model fit for a parameter of a structure. The first optical model fit is based on a domain of quantities for a first model of the structure. A first near optical field response is determined for a first quantity of the domain of quantities and a second near optical field response is determined for a second, different quantity of the domain of quantities. The first and second near optical field responses are compared to locate a common region of high optical field intensity for the parameter of the structure. The first model of the structure is modified to provide a second, different model of the structure. A second, different optical model fit is determined for the parameter of the structure based on the second model of the structure.

Abstract: Methods and apparatuses for calibrating attitude-independent parameters of a 3-D magnetometer are provided. A calibration method includes storing and updating data related to a N×9 matrix T and a N×1 matrix U extended for each measurement with an additional row and an additional element, respectively, the additional row and the additional element being calculated based on values measured by the 3-D magnetometer for the respective measurement. The method further includes calculating analytically (1) a symmetric non-orthogonal 3×3 matrix D representing scaling and skew of the 3-D magnetometer measured values and (2) a vector b representing bias of the 3-D magnetometer measured values, using the stored data and a singular value decomposition (SVD) method.

Abstract: In some aspects, a method for performing analog matrix inversion on a matrix with a network of resistive device arrays B, W, Q, and C is described. The method may include initializing arrays W, Q, B and C, updating the connections of array W in parallel and array Q in parallel until a predetermined condition is satisfied, and responsive to determining that the predetermined condition is satisfied, outputting an inverted matrix based on outputs from the connections of arrays B, W, Q, and C.

Abstract: A method, computer program product, and circuit are provided for noise and bound management for a Resistive Processing Unit (RN) array having an op-amp. The method includes reducing the noise in an output signal from the RPU array by using a largest value, in a sigma vector having a plurality of values, as a representation for a window for an input signal to the RPU array. The input signal to the RPU array is formed from the plurality of values. The method further includes sensing saturation at an output of the op-amp. The method also includes managing the bound to eliminate the saturation by reducing the plurality of values from which the input sign to the RPU is formed.

Abstract: Techniques for estimated compound probability distribution are described herein. Embodiments may include receiving a compound model specification comprising a frequency model and a severity model, the compound model specification including a model error comprising a frequency model error and a severity model error, and determining a number of frequency models and severity models to generate based on the received number of models to generate. Embodiments include generating a plurality of frequency models through perturbation of the frequency model according to the frequency model error, and generating a plurality of severity models through perturbation of the severity model according to the severity model error.

Abstract: A page component dynamic layout method includes: initiating, by a client terminal, a page content request to a server based on a triggered request page; querying, by the client terminal, whether there is a preset identifier in a page response content; if there is a preset identifier in the page response content, loading a set custom component parsing engine to parse a tag of a custom component; selecting, by the client terminal, a corresponding custom component according to tag information obtained by parsing the tag of the custom component, and calling a set custom component rendering engine to render the custom component; and adding the rendered custom component to a corresponding position of the request page. Dynamic layout of a Native component on a WebView is achieved on an HTML page, thereby improving the convenience of laying out pages by an operator.

Abstract: Collating text strings having Unicode encoding includes receiving two text strings S=s1s2 . . . s and T=t1t2 . . . tm. When the two text strings are not identical, there is a smallest positive integer p for which the two text strings differ. The process looks up the characters sp and tp in a predefined lookup table. If either of these characters is missing from the lookup table, the collation of the text strings is determined using the standard Unicode comparison of the text strings spsp+1 . . . sn and tptp+1 . . . tm. Otherwise, the lookup table assigns weights vp and wp for the characters sp and tp. When vp?wp, these weights define the collation order of the strings S and T. When vp=wp, the collation of S and T is determined recursively using the suffix strings sp+1 . . . sn and tp+1 . . . tm.

Abstract: Systems, devices and methods operative for identifying a reference within a figure and an identifier in a text associated with the figure, the reference referring to an element depicted in the figure, the reference corresponding to the identifier, the identifier identifying the element in the text, placing the identifier on the figure at a distance from the reference, the identifier visually associated with the reference upon the placing, the placing of the identifier on the figure is irrespective of the distance between the identifier and the reference.

Abstract: In one set of embodiments, a web document is associated with a style sheet that can include style rules, which provide details about how the document is to be presented and formatted by, for example, a web browser. A reduced set of style rules is determined based upon contents of the web document and style sheet. Reducing the size of the style sheet is desirable, e.g., to increase communication and processing efficiency when the style sheet is sent with the document in an e-mail message. The reduced set of style rules can include rules that are referenced by user interface components present in the document, and can exclude rules that are not referenced by the document, or are incompatible with or not applicable to e-mail. The reduced set of style rules can be embedded in an e-mail message or other communication in place of the original style sheet.

Abstract: Systems and methods for optimizing space utilization in a document are described. The system receives a template divided into static space, dynamic space, and whitespace. The system fetches static contents, dynamic contents, personalized contents. The system determines the sizes of the contents and spaces available in the template. Further, the system compares the sizes of the contents with their respective spaces of the template. Based on the comparison, the system may resize at least one of the contents and the spaces relative to each other. Based on the comparison, the system may populate the contents inside the spaces without compromising with resolution and compatibility attributes associated with the contents. Post populating of the contents, the system may generate a final document for user.

Abstract: In various implementations, a computing device: displays the document outline on the user interface; maintains, in a non-transitory computer-readable medium, a causal graph data structure representing the document outline, wherein the causal graph data structure includes a plurality of structure nodes, each structure node representing a level of the document outline; receives, via the user interface, an insertion of a new level to the document outline; in response to the insertion, defines a structure node that represents the inserted level; inserts a transition node in the causal graph data structure, wherein the transition node represents a relationship between the structure node and at least one other node of the causal graph data structure; inserts the structure node into the causal graph data structure as a child of the transition node; and updating the user interface to display the inserted level.