KR20020075359A - System and method for capturing and managing information from digital source - Google Patents

Abstract

Systems and methods are provided that allow a user to capture, in addition to other items, items of individual information, including files that are of interest to the user or that represent auxiliary components of the information from the document. In addition, the context database is captured for a user that includes metadata including captured items, files associated with the items, and keywords associated with the item. The context database is queried for the creation of the context summary in response to selections by the user or words entered by the user. The context summary is used to enhance the search and select target advertisements for display to the user based on the user's current activity information.

Description

System and method for capturing and managing information from digital source

(Explanation of Background Art)

The excellence of the World Wide Web as an important source of information has enabled end-users and product developers to e-mail, web pages, And exploring how to effectively capture useful information from digital sources, such as documents.

One basic classification of capture techniques is standalone capture techniques. For example, if the user wants to save a paragraph on a web page, the user can print the contents of the web page, save the web page object as HTML or text files, and then use Microsoft Word or PowerPoint, The product can be used to copy and append web page content to a "carrier" document, or simply write the web page content of interest on the paper. Another capture technique captures and manages the entire contents of web pages or sites in most browsers that capture and manage links to interesting sites that have been visited, and files on local disk for subsequent browsing. Browser-based page capture, including Favorites and Bookmarks tools available in offline browsers (also known as "page grabbers") tools).

All standalone capture techniques have serious limitations. Hard copies have very limited options for aggregation or presentation and are easily damaged or broken. Capturing web objects in files is tedious and time consuming. Although many objects logically belong together, typically only a single object is stored per file. Objects typically have to be managed with limited utility of standard file systems (eg, Windows Explorer), and aggregating objects in files requires manual pasting into carrier documents.

The use of carrier documents in web objects is extremely difficult, and system resource-intentive often causes subtle or even rapid changes in the appearance of captured objects and stores one object per document under the document management system. If no unique means is used, the use of carrier documents provides only limited coarse-grained management and retrieval performance.

Browser based page capture tools also have serious limitations. Capturing links to web sites and web pages or capturing the entire contents of these sites and pages is by definition a crude grainy behavior, and thus heterogeneous information is accessed, and the user organizes the captured content. Can not control. Page capture tools (link or content based) typically use only name strings, URLs, and structure information for classification and management. Additional sorting tools, such as keywords and notes, are usually crude or statically aggregated pages containing many items that cannot be described in detail in meaningful terms. .

The crude grainy nature of page or site-based management and the lack of complex classification tools bother the process of placing information within an extensive favorites list or offline page storage area, ie effectively limiting the size at which this usefulness can increase. .

Pure link-based site or page capture does not provide secure or guaranteed accurate results because the content of addressed pages is changed without notice, links become stale and dynamic pages disappear after they are addressed. .

To date, businesses committed to e-commerce generally have high traffic, such as search portals for collecting promoters that target information about customers and for actually delivering advertising. -traffic) have been dependent on web sites. These high traffic sites typically get customer interests and purchase patterns through two basic techniques. First, sites implicitly collect information through tracking "click trails" to determine which sites a user visits and what the user links to. Next, sites offer promotions, contests, and site personalizations where heads up voluntarily provide more information about themselves in exchange for additional services, greater online convenience, or opportunities for thought. Collect information publicly through its characteristics.

The main limitation was the inability of current Internet information gathering technologies to provide specific actionable information about consumers. Even a portal site has "presonalization" because it requires the user to set up personalized pages that yield "opt in" and a limited amount of user data.

The ability to provide context, some underlying, or background information, is highly desirable to assist search engines in finding the desired information. For example, a user may find information about handling elephants in Africa, but a query may include numerous results about handling elephants in circus.

It is highly desirable to provide fine grained access and storage of individual information that can also be used to provide context for search and advertising that suits the needs or needs of the user.

(Related application)

This application is incorporated herein by reference, and as a "Provisisional Application" filed on July 2, 1999, the inventors David S. Wadhwani, Dennis S. Buchheim, Richard S. Buchheim, Scott A. US, entitled "System and Method for Capturing and Managing Information form Digital Sources," by Raposa and Ethan F. Malasky. Provisional Patent Application No. 60 / 142,237 to 35 U.S.C. Claim priority under δ119 (e).

(Background of invention)

The present invention provides a system and method for capturing and managing information from digital sources.

1 illustrates one embodiment of a system for capturing and managing individual information from a digital source in accordance with the present invention.

2A illustrates an embodiment of a user interface used in a system for capturing individual information in accordance with the present invention.

FIG. 2B illustrates an embodiment of a collection of individual items of a collection represented by a browser display. FIG.

FIG. 2C illustrates a pop-up that allows a user to indicate whether the selected item is to be saved as an individual form of the item by selecting "web content" for that item, or to store the link on the web page where the item appears by selecting "web link". Figure showing an embodiment of a pop-up window.

FIG. 2D illustrates an embodiment of a user interface, pop-up window in which metadata is displayed to a user and accepts additions or changes to metadata from that user. FIG.

FIG. 2E illustrates an embodiment of a user interface as a pop-up window displayed to a user to allow annotation of an item. FIG.

FIG. 2F illustrates an embodiment of a user interface as a pop-up window displayed to a user to change the appearance of an item. FIG.

FIG. 3A illustrates an embodiment of a computer implemented system for capturing and managing crude grainy information and individual items of information from digital sources as implemented in a browser environment.

FIG. 3B illustrates the implementation of a context database as a WebBase containing files for each item captured and containing metadata stored as SQL database entries and files stored in a hierarchy of collections, according to table definitions. A diagram showing an example.

FIG. 3C illustrates an embodiment of a user interface, pop-up window, to allow a user to create a filter of a user context database. FIG.

3D illustrates an embodiment of a user interface, pop-up window, in which filters to use in a collection are displayed.

4 is a flowchart of an embodiment of an overall method of forming an item in a uniform file format.

5A and 5B show an overall flow diagram of one embodiment of a process of generating metadata for an item.

6A illustrates an embodiment of an item access system as a file system hierarchy for items and objects.

FIG. 6B illustrates an example of how three items with references to external files are stored in the file system embodiment of FIG. 6A.

7A is a flow diagram of an embodiment of a method for creating a context summary based on an embodiment of a context database when a user selects an item or collection.

7B illustrates an example of using the method of FIG. 7A.

8A is a flowchart of an embodiment of a method of selecting a target advertisement using a context summary in accordance with the present invention.

8B is a flow diagram of an embodiment of an improved search method.

9A illustrates a user's selection of an embodiment of a user interface.

FIG. 9B illustrates an example of an improved search query and suggested keywords generated from the method of FIG. 8B.

FIG. 9C illustrates an embodiment of a web page that includes a targeted advertisement selected from the method of FIG. 8A.

(Summary of invention)

The present invention provides a system and method for capturing and managing information from digital sources.

In one aspect, the present invention allows a user to capture and manage individual information from a digital source, such as a web site, email or electronic document. In general, in order to capture individual information, such as paragraphs, from a web site or other digital source, the user may highlight the desired information on the web site and display it on a user interface such as a browser indicating that the highlighted information should be captured. Enable capture control. In another aspect of the invention, the captured information referred to below as an item is advantageously a uniform file format, such as enhanced HTML / XML, that allows for retrieval of captured information and fine grained access to the captured information. format). Each item is stored in an item access system. One embodiment of an item access system is a file system in which items are stored in files. Items are also analyzed and organized by assigning metadata to captured information. Examples of such metadata include the name, category, keywords, source, source description, capture date and time of the captured information. The user can select metadata for the captured item proposed by the present invention or alternative, and the user can enter the user's own metadata.

The present invention also manages items of individual information using a context database that includes items, any related files, and metadata associated with the item. Once an item of information is captured and stored in a file, its metadata is integrated into a database, such as a SQL database, to assist in the classification, annotation, structure, and management of the captured information. The context database allows the user to dynamically gather the captured information. In other words, the user can combine or recognize different captured information items. One embodiment of a context database is WebBase, where items are stored in a markup language such as HTML / XML so that the items are easily managed over the Internet. The web base embodiment of the present invention brings together the power of a database or the ubiquity of web pages. This melding of web pages and SQL database enables peripheral features such as sharing and import / export in many small additional tasks.

In another aspect, the present invention uses a context database to convey targeted promotions, services, and content for a product and to assist the user in finding additional information similar to information already captured. In general, the present invention relates to metadata stored in relation to each captured item, items stored in a context database to create summaries of particular user interests at any point in time, and relationships between the items. Analyze the relationships between the metadata of items. The resulting summary can be analyzed by a computer and then used to provide targeted timely promotional links to specific products, services, and content when a user retrieves an item from a web base. Provides an overview of user interests in a format.

In another aspect, the present invention uses metadata associated with captured information in a context database to enhance searches performed on digital information systems such as the Internet. In general, when a user starts a search, the present invention analyzes a summary of user interests appropriate for the search to determine which words to use in the search and how to use the words in the query. Thus, the metadata associated with the captured information is used to make a powerful search query that adds information from the user summary of interest that can be passed along existing search portals.

Thus, the present invention advantageously allows a user to capture and manage individual information from digital sources.

The accompanying drawings show embodiments of the invention. The description of the embodiments of the invention is provided for the purpose of illustration and description, and is not intended to be exhaustive or to limit the invention to the precise form disclosed. Many modifications and variations are possible in the art, and the scope of the invention is not limited by this detailed description, but is intended to be limited by the appended claims.

1 shows a customer computer 118 'with an operating system 164', a memory area including a storage area for a user's context database 140 ', and a display with a user interface 160' such as a browser display. 108 '), an embodiment of a server 128 that communicates 142' with a customer computer 118 'via input devices (keyboard 114 and mouse 116), a network 159' such as the Internet. The server hosts a website 130 that communicates with the customer computer 118 ', and a storage area 138 to provide alternative storage locations for the plurality of context databases 140', and the website 144 and digital sources such as e-mail or electronic document 148 also capture and manage individual information from a digital source in accordance with the present invention, including an embodiment 150,154 in communication with a customer computer 118 '. One embodiment of a system is shown. Another embodiment of a guest computer is also shown in FIG. 1 as a personal digital assistant 118 'having a stylus 126 as a display 108' and a user input device, and an operating system 164 '. The Personal Digital Assistant (PDA) also communicates with the server 128 and the network 159 '158. Figure 3A shows that the operating system drags and drops the clipboard structure 168 and the drag and drop infrastructure. infrastructure: 170).

PDA embodiment 118 'also has a user interface, such as a browser display 160', for presenting information to a user. Browser display 160 'is a capture control implemented as a toolbar having three click buttons 120', 122, 124 'and an item access window 162 for displaying items in a user's context database. have capture control. In these embodiments, the brightest click button 120 'activates the brightest look for selecting interesting information. One click capture button 122 'is activated by the user to indicate that information should be captured. Third click button 124 ′ is shown in these examples to allow the user to indicate whether the user's individual items are invisible.

In FIG. 1, each display is a user selection of an individual item of information 112 ′ that is selected in response to user actions of an input device, such as mouse 116 and needle 126, depending on the drop and drain infrastructure of the operating system. (110). The keyboard 114 may also be used as well as any other input devices for selecting the desired individual item of information. The user can also click on the brightest button 124 'and use the input device to move the cursor over the individual information to be selected. The user then clicks the one-click capture button 122 'to indicate to capture the selected item.

2A illustrates an embodiment of a browser display 200 generated by one version of the present invention. The browser display 200 has a toolbar 202, a browser window 204, an advertising panel 206, a browser plug-in window 222, and a browser plug-in window ( 222 includes a search word entry bar 208 and a collection map 212 for displaying the user's items in the structure of the user's context database, which in this embodiment is a web base, and also includes a file menu button ( 214, one click capture button 216, send button 218, web search button 220, context database search button 227, and tools button 225, and item list 210 ).

As shown, a collection titled PTO includes a number of items, including two items of individual information, each of which is referred to as "Intellectual Property Symposium of the Americas" (228) and " Stored in a file identified by the names of some well-known US Trademarks " The browser window 204 displays the contents 215 of the selected item “US Intellectual Property Symposium” 228. The advertising panel 206 also displays the displayed webpage 200 for displaying related advertisements and promotions as selected in one embodiment, based on matches of keywords of various advertisements and promotions and keywords of the user's context database. Included).

2B illustrates an aspect of the invention in which individual items are aggregated for a user. When the "PTO" collection 226 is selected in this embodiment, the browser window 204 displays the individual items in the collection in one aggregated format display 217. In one embodiment of the present invention, the items may be dynamically collected, for example, FIG. 2B shows the "US Intellectual Property Symposium" 228 that is displayed together when two items, PTO collections, are selected. "Some well-known American brands celebrate 100 years" (230). The captured items of information are processed so that previously anonymous documents can be intelligently integrated. This facilitates browsing, sharing, exporting, importing, searching and transmitting collections of individual items related to others. As will be appreciated by those skilled in the art, many other forms of dynamic aggregation of webbase items are possible. The ability to perform a dynamic aggregation means that information can be easily and easily reorganized and reconsidered in a way that simply cannot be made possible with traditional documents based on systems.

Multiple and often mixed types of formats of digital information can be captured by acquisition control. Mixed types of formats include links and other files, such as Word and excellent documents, as well as third-part insertable objects such as movies, applets, flash, shockwaves, tests, etc. It is not limited to images, sounds.

In the capturing items, users have the option of capturing crude grain information or fine grain information. Examples of crude grain information are web pages, documents or encompassing files. An example of fine grain information is a sub-component of a document or include file. In one embodiment commonly used with respect to capturing information in the World Wide Web format as shown in FIG. 2C, a user may enter an entry by selecting the "Web content" 258 for that item. It may be saved in a separate form that is smaller than the document's data, or may indicate whether the item is typically saved to a web page that appears by selecting “Web Link” 260. In this embodiment, a name (262) is provided for the item, a notes section (264) and a folder (226) are provided for that item, all of which are changed or added by the user. Can be.

As shown in FIG. 2D, a pop-up window 299 allows a user to click scroll bars 248 and add (291), manage 293, and OK 253. The user is provided to provide a name for item 242, category 244, and suggestions for more keywords 2486 ′ that can be selected by manipulating the buttons. Provided to the user in the window 257 of the box are the source of the item, a description of the source and the date and time of the capture. The user clicks the OK button 253 to get the item to be inserted into his context database. The user can also change the appearance of the item by adding a notification or clicking on each of the 'notes' 292 and the Appearance 294 buttons. The example pop-up window 252 shown in FIG. 2E is indicated to allow the user to enter an entry of notifications, and the example pop-up window shown in FIG. 2F can be used to change the appearance of an item. It became.

3A shows an embodiment of a computer implemented system 320 for capturing and managing individual information from digital sources inserted as a browser 360 operating on a customer computer 118 ′. The user's selection of the one click capture button 120 'or 216 is communicated with the capture control 322. Acquisition control 322 interacts with operating system 164 and application specific access protocols. For example, when a user selects 110 'an individual item of information 112' and activates a one-click capture button as shown in FIG. 1, the item is passed to the operating system clipboard. In the illustrated embodiment, capture control 322 examines the content captured on clipboard 168 in the available clipboard formats, and selects one that provides the best translation for Uniform File Format. do. The item in the selected format is then passed along the uniform file formatter 324. In another example, when a user captures an item directly from the browser 360, the capture control 322 examines the browser's internal data structure or document object model and associates it with this data set in the uniform file formatter 324. Pass along the formatting information.

The uniform file formatter 324 places the captured items in a uniform format. In one embodiment of the present invention, the uniform file formatter transforms all items into an "aggretabable" format so that they can be gathered dynamically. For example, FIG. 2B shows items in the selected POL collection 226 aggregated into one displayed HTML document 217 in the browser window 204. Items stored in an aggregateable format maintain presentation integrity regardless of their position in the aggregated display.

Although the metadata generator 326 is shown as a separate element, its functions and the functions of the uniform file formatter 324 may be performed together in a single pass of the captured item. Meta data generator 326 associates the item with metadata. Examples of metadata include, without limitation, name, category, keywords, source, source description, and the like. Meta data is used to provide context to the user's context database 140 '. The generated metadata is provided to the user via the user interface as shown in FIG. 2D for approval and / or modification and addition. The metadata generator includes their changes and / or additions received from user 338. The metadata generator delivers the metadata to the context database manager 330 so that the metadata can be stored in the user's context database 140 '. Moreover, uniform file formatter 324 and metadata generator 326 are in communication with generated file manager 328.

The generated file manager 328 will store the items in a collectable format in the file system 354. In one embodiment, when an item or collection is viewed, the generated file manager dynamically generates a header that is written to the buffer. The generated file manager places the item or items in a "gatherable" format when the collection is selected, and appends the item or items to this header. The resulting buffer is written to a file and displayed to the user.

Context database manager 330 stores, corrects, and retrieves requested data for other components of the system from context database 140 '. The context database includes a pointer to a location in memory of a user's formatted files 354, a database 356, and other characteristics, such as a collectible item associated with the user's items, and seed keywords ( Seed Keywords: 358). Examples of seed keywords can be found in Appendix C of the Provisional Application. In one embodiment of the invention, the seed keywords 358 are used by the context summerizer 332 under certain conditions in making the context summary.

The context summerizer 332 makes a context summary that is a weighted, keyword-based summary of the user's context database 140 by querying the database. The context summerizer 332 includes a context filter 336 that prioritizes keywords in the context summary based on the currently selected item or collection. The context summerizer creates a context summary every hour, and the user selects an item or collection and passes the summary to a profile analyzer (342) located in the server 128 in this embodiment. The context summerizer also creates a context summary of keywords when requested by a Search Enhancer that returns the summary.

Profile analyzer 342 receives a collection summary according to each selection of an item or collection. In one embodiment, the context summerizer 332 includes a category of items. In another embodiment, profile analyzer 342 obtains the category of the selected item or collection by requesting it from context database manager 330. The profile analyzer then forwards the summary and category for the current selection to a promoter interface 344 located on server 128. The promotion interface selects a promotion to be displayed to the user. In one embodiment, the promoter interface compares the keywords of the current selection with keywords 345 associated with the promotion 343 stored on the server to find matches. If a match is found for the keywords, those promoters that match the keywords are selected. In the embodiment of FIG. 3A, the promoter interface generates an HTML document for display in the advertising panel 206.

Search Enhancer 346 analyzes a user's search string entered in search query entry bar 208. The search enhancer forwards a request to the context database manager 330 to search the context database to find occurrences of words in the metadata and to search for items or collections related to those words. If any items or collections are associated with the words in the search string, the search enhancer requests a context summary for each item or collection from the context summer.

The search enhancer creates a search summary from the context summers that are passed according to the user's selected search engine to the query generator 348 on the server 128 in this embodiment. The query generator 339 places the words of the search summary in place from the Search Engine Grammar Protocals 339.

3B illustrates an embodiment of a context database as a web database 370 that includes an SQL database 376 and users' files 378. The user's files include the user's items 372 as well as associated files 374 including files for referenced objects 374 such as HTML frames or embedded objects such as images or movies.

Users of the SQL database are helped in classifying, annotating, organizing, and managing the captured information. An SQL database contains tables that define relationships in order to provide a context for the user in managing a Web database. Examples of table definitions are included in Appendix B of the interim application, which is incorporated herein by reference. In particular, the tables include object table 308, relationship table 382, filter table 392, filter relationship table 394, information map table 388, information map expkeys table 390, tag table 386, and a tag relationship table 384. In one embodiment, the item may be represented by an entry in the object table 380. Columns are keyed in order of a particular item, its name, notifications, source, source description, pointers to the location of the file, appearance characteristics such as background color, and child relative to the parent collection. Include them without being limited to keys for keying. Each item belongs to one or more collections as defined in relationship table 382. For example, each item may be related as a child of one or more parent collections, and each collection may in turn be related as a child of one or more parent collections.

In this embodiment, the keywords, categories, and names for each item are represented by entries in the tag table 386 where the keywords, categories, and names for each item are related to the items in the object table by the relationships defined in the tag relationship table 384. Can be.

In this embodiment, the information table 388 is defined to maintain the user's selection and the current state of what the user is viewing. The information map Exp keys track the current open tree path by interacting with relationship table 382 and recognizing that collections and items are open. When the system is running, this information in random access memory (RAM) can be easily obtained from RAM rather than performing database queries on information map tables.

Filter 392 and filter relationship 394 tables assist the user in sifting through the user's context database. The filter table 392 contains various rows that define fields such as desired rows, a filter or agent name, a comment, and a key for indexing with related keywords. The filter relationship table 392 associates items in the object table according to the selected filter with the filter.

For example, FIG. 3C shows a pop-up window 395 as a user interface entitled "Agent Properties" that allows a user to define a filter called an agent in this embodiment. This embodiment In the filter table 392 is the name of the agent, the name of the item, the keywords of the item, the notifications of the item, and the words of the entries in the entry bars 415, 401, 393, 403, and 405 related to the source of the item. The columns corresponding to the user entry In the example of Fig. 3c, the user has entered a "patent" in the keywords entry bar 393. This means that the context database has a filter and the word "patent" associated with them. Create a filter relationship column for this agent “patent” that associates items or collections that have a PTO collection, as shown in pop-up window 407 of FIG. 3D. By clicking on button 417, "government" agent 411 can also be used for PTO collections.When a filter command is received from a user, these items or collections associated with the filter are received by the user. Is displayed.

Before the captured item of information is stored in the user's context database, the item is formatted in a uniform file format. 4 is a flowchart of an embodiment of an overall method of formatting an item in a uniform file format. 5A and 5B illustrate an embodiment of a process of generating metadata. In one embodiment of the present invention, a single pass through the captured item is effectively used for the purposes of a uniform file to format and generate metadata.

A single format that can include many other formats is highly desirable as a uniform file format. Examples of formats that may come across include markup languages such as HTML and XML, URL to ASCII files (including local files), rich text format, embedded files and hyperlinks. Property binary formats, such as those used with some word processors for hyperlinks, local binary files such as storing images and other multimedia media, URL to binary files, and data. "Self-managed" embedded objects such as registered components that are interpreted and displayed and mixed with other data such as Microsoft Excel, Microsoft Word, Acrobat PDF, Flash, and ASCII text. And references to these.

HTML is used in the examples below as a uniform format. If all the conversions are needed, it is repeated to ensure that the referenced files are properly converted.

4 illustrates an embodiment of a method for converting an item of information into a uniformly aggregateable HTML file format. First, the format of the item is determined (402). In the example of FIG. 3A, acquisition control 322 provided the format. Next, a local representation of the current item is generated in the HTML (404). In this embodiment, file formatting and metadata generation are done together in a single pass of the captured item. In preparation for the embodiment of the metadata as shown in FIGS. 5A and 5B, a name suggestion buffer and a keyword suggestion buffer are prepared 418.

If the character set used by the item is not a single uniform character set (420), the characters of the item are converted to a single uniform character set (422). Examples of common character sets are Unicode or common unicode mapping (UTF8). This transformation is an example of a "settable" variant. This allows a set of two items that were previously incompatible based on property character set formats.

Next, the existence of references to other external files is determined 406. If references to external files exist, the referenced files are downloaded 408. The downloaded files are made of interchangeable HTML. For example, IMG tags are located around the images, and anchor tags are located around the links. Also, for external files, references in the item are determined to point to local copies of the files (412).

The file is located in aggregable format 414, for example, by transforming objects that are absolutely positioned relative to objects that are located relatively. This allows the item to maintain its presentation integrity regardless of whether the item falls on the aggregated display. In addition, the header information of the item may be removed from the item's HTML (if present) and stored in the context database. An example of a property that may be saved for presentation integrity in header information is the front or colors in which portions of the items are displayed. In the embodiment of FIG. 4, pointers to files in a uniform collectible format are stored 416 as metadata containing the characteristics of the item for subsequent header generation upon dynamic aggregation.

The examples below are provided to illustrate formatting items for a uniform format.

The captured item may already be in the version of HTML. Local representation of HTML is generated in the version used as locally stored HTML. In general, HTML often refers to external files to be inserted into a viewable document. The referenced information is downloaded and analyzed to point to local copies. Again, the stage is managed repeatedly to get all the referenced files and their referenced files. In one embodiment, Java Script manages all references with a consistent and unique prefix. The following is an example of how structures can be converted. Frames can be called for IFRAMEs, alternatively TABLEs, to see the combined frames captured on pre-HTML 4.0 systems. Again, an iterative approach is used to make frames acceptable within frames. Images, sound files, movies can be stored in local or remote image files. The file is downloaded if necessary, and the references are made as local copies.

To capture the referenced Java applets, the applets are downloaded and analyzed to reference a local copy. The capture of the applet is initiated by the capture of the remaining HTML file.

Some captured items or referenced files have parameters that can be entered. For example, in the case of a major third part, such as a dictionary of Flash or ThingWorld, the data file of the embedded application is downloaded as well as the files referenced from the parameter list. References to local copying must be analyzed and parameters changed so that the content can be displayed without disturbing information, formatting, or function. Flash allows flash developers to rely on arbitrary parameters such as input files. The flash file is downloaded, and this input file must also be downloaded. The parameters inside the flash tags need to be changed to refer to the newly downloaded version.

In the example of an embedded application that is not recognized, the application can be downloaded or executed. Heuristic teaching is needed to analyze the parameter list by analyzing the values. If it has an extension with "/" and ".", We assume that it is a reference to the file being downloaded. The reference is determined to be a local copy, the parameters are changed, and the content can thus be displayed without disturbing information, formatting or functionality.

With regard to capturing URLs for text files that are translated into markup languages such as HTML or XML, the following example provides other options. URLs are simple references to files (HTML). In one embodiment, the user is given two options. First, the file being referenced is captured, which is simply restarted to handle data such as file types rather than URLs.

Another option is to only capture URLs as bookmarks. The URL is then wrapped in appropriate HTML / XML reference tags, such as <A HREF> </A>.

In file conversion from Rich Text Format (RTF) to HTML / XML, one approach begins by analyzing the RTF and converting HTML tags.

Patent binary formats are encountered by chance, the uniform file format is HTML / XML, and these formats cannot normally be inserted into an HTML viewer. The file is treated as an HTML link, and references to the file are wrapped in <A HREF> </A> tags.

If a local binary file is encapsulated in HTML / XML, the file type is analyzed taking into account file header information, file suffixes, and operating system registration suffixes. Based on industry standard suffixes later, the file is not limited to "IMG" tags for video, "SOUND" tags for sound files, or "EMBED" or "OBJECT" tags for recognized HTML / XML plug-ins. Wrapped in the appropriate HTML tag.

If the uniform file formatter encounters a URL to a remote binary file to be converted to HTML / XML, the remote file can be downloaded and managed as in the case of a local binary file.

References to "self-managed" inserted objects to be converted to HTML may request to wrap the object in an "OBJECT" tag with appropriate parameters as determined by a query to the operating system. If it cannot be achieved from the operating system, certain dictionaries of objects that can be inserted later are considered.

In converting ASCII text into HTML / XML, the text is passed through a conversion filter. Formatting is done to maintain the integrity of the text. For example, spaces are converted to "& nbsp", CRLFs are converted to "<BR>", "." Is converted to "& gt", and "<" is converted to "& lt".

HTML has rules and restrictions that the contents of the text can trigger or violate. When text is viewed as HTML, an interpreter (like a web browser) is passed along a transform filter that outputs HTML that displays the captured content.

5A and 5B show an overall flow diagram of one embodiment of a process for generating metadata for an item as may be done by the metadata generator. The metadata is associated with all items that the user captures, and as shown in FIG. 2D, this metadata may be viewed or changed by the user. In one aspect of the invention, the generated metadata includes name, category, keywords, source and source description for each item.

A name suggestion buffer and a keyword suggestion buffer are prepared (418). The content of the capture item has been filtered to provide "appropriate" information, meaning that the formatting tags (bold, italic, position, etc.) have been removed. Appropriate information was stored in butters. Examples of suitable information include any descriptive tags, such as the "ALT" tag of HTML that is considered "appropriate" for the proposed name of the item and is retained in this example. This buffer may be limited to a certain size, such as 5K, in order not to negatively affect the capture performance of the product.

The metadata generator makes the proposed name of the first N, for example the preselected number that can be determined by the file system is such that the combined length including the spaces is less than a predetermined number of characters 510, eg less than 128. Limit the whole words. Once the proposed name is determined, it is checked for an improperly close construct (512). If the proposed name ends with an improperly close configuration, the last word of that name is removed 514 repeatedly until the close configuration is no longer inappropriate. In this case, the proposed name is selected as the name of the item (516). Below are examples of inappropriate configurations.

"the", "this", "and", "an", "that", "for", "to", "or", "nor", "though", "although", "if", "who "," which "," whoever "," what "," whose "," these "," those "," whether "," but "," so "," yes "," as "," because ", "until", "when", "whenever", "even", "rather", "about", "at", "against", "by", "during", "except", "from", "in "," of "," like "," till "," upon "," with "," while "," since "," & "," ... "," | ",": ","; ",", "," / "," ＼ ","-","-"," = "," + "

As will be apparent to those skilled in the art, other words may be selected for removal and addition from this list.

In the embodiment shown in FIG. 2A, when items are captured, a category is assigned to something like "Business" or "Travel". First, an item is placed in a collection along with a category, which is an "appropriate" context collection in which the item automatically inherits its category if one is not already selected by the user yet. 5B shows that a category of items is assigned based on the first " appropriate " context collection associated (518).

5A and 5B, the source of the item is associated with the item based on information received from the source application during data delivery (520).

In one embodiment, the metadata portion of the item is a description. The description may also be made from a name suggestion buffer of appropriate information. For example, if the uniform file format is in a resolvable markup language 522 and a title tag 524 is present, the description of the item or the source description is set to the content in the title tag 526. If there is no title tag associated with items 522 and 524, then the description of the item may be set to its source (528).

Next, the keywords are associated with the item. 5A and 5B, the appropriate information is not only stored in the name suggestion buffer, but also in the keyword suggestion buffer (506). In this embodiment, a larger keyword buffer 50K is required.

In determining the keywords, the metadata generator is drawn on the keywords from one or more below sets of keywords. The first set is User Keywords that refer to the set of all user keywords rather than the context specific set described as context keywords. As shown in the example of FIG. 2D, the user may enter keywords to be associated with the item and take suggested keywords for the item. These keywords form a set of user keywords.

Another set of keywords is a set of Seed Keywords that includes a database of sheet keywords 358 that can be particularly helpful before users set up their own useful database of keywords. These keywords are grouped by category to allow smart keyword suggestions.

There is also a set of context keywords. Context keywords are those keywords associated with items and collections in the path of the item from the top of the hierarchical collection to the item. The path has many levels of collections or no levels. Limits on the number of levels to be searched (eg 3 levels) and minimum and maximum number of keywords may be required.

An example of a set of context keywords is

1. All keywords assigned to a collection in the path from the top of the hierarchical collections, such as "All Collections" in FIG. 2A, for example a designated collection such as the "PTO" collection in FIG.

2. All keywords related to items in the specified collection. In FIG. 2A, this includes keywords related to "Intellectual Property Symposium of the Americas" and "Some Well-Known U.S. Trademarks."

3. A combination of keywords related to items in collections in the collection path.

5B continues the generation of metadata including the generation of keywords. In this particular example, it is first determined whether the capture specification of the item has a category assigned to it (530). If a decision is made, the SQU query of the user's context database, e.g., Object (380), Relationship (382), Tag (386), Tag Relationship (384) tables in the database is a combination of the first three sets. It is made to capture all keywords that satisfy the inset. The result from the query is a set of context keywords for a particular item.

Moreover, the description collection has a category, and seed keywords associated with the category can also be analyzed for matches.

In the illustrated embodiment, the minimum number of keywords found, such as a keyword threshold of 3, is also used as the maximum keyword suggestions, which is a parameter that determines the maximum number of keywords to be proposed for the item.

The metadata generator loops 532 through the set of sheet words and context keywords associated with the category of the specified collection that matches each set for captured content. The loop is bounded by keyword thresholds and maximum keyword suggestions (532). Also, matches made for the earlier set (s) are never replaced by matches from later sets. The keyword threshold is not satisfied after moving through all these keywords (536), and then the selection of possible keywords for matching is broadened if the specified collection does not have a category. Keywords, on the other hand, are assigned to an item as tracked by metadata in the user's context database.

If the specified collection of this item does not have a category associated with it, the keyword sets include context keywords, all user keywords, and all sheet keywords. Again, the metadata generator loops 534 through a set of context keywords, all entered user keywords, and all seed keywords that match each set to captured content. The keywords obtained are then assigned to the user's web base 538. In the embodiment of FIG. 3B, the metadata generator delivers the keywords obtained with a web site manager incorporating these keywords with a user web base as defined by tables in the web base.

6A shows an example of an item access system, a file system hierarchy for storing items, and the objects that it references. In this example, the top directory is the Installation Directory (602). The next lower level is a data directory 604 that includes item HTML files 602, generated display files 612, and an item related directory 608 that is at the same level as the item HTML file. The item related directory 608 is files related to the directory 610 that mean the objects referenced by the related item file.

FIG. 6B shows an example of three items formatted in HTML that refer to external files using the file system hierarchy of FIG. 6A.

In FIG. 6B, main directory 614 is the topmost directory in the file system hierarchy. The first item 618 was captured from the web site and recorded in the file system as 123.htm 618 in the data directory 616. The referenced file is also downloaded to a directory called 123 624, which is at the same level as 123 htm in the file system hierarchy. When HTML is processed for this capture, the appropriate references are changed from the references to the 123 directory 624 from the references to the World Wide Web.

Similarly, the second capture item is captured from another web page and stored under data directory 616 as 234.htm 620, the referenced files for 234 are downloaded and at the same level as 234.htm in the file system hierarchy. In directory 234 630. When the file marker Maker processes this item, the appropriate references are changed from the references to the 234 directory 624 from the references to the world wide web.

The third item is a text file captured as a copy from the file system. In this embodiment, the uniform file format is HTML and this text file has been converted to HTML. The HTML file is saved under the data directory 616 as 345.htm 622, and the text file 640 in the file system hierarchy. It is stored in a directory called 345 (638) at the same level as .htm (622).

For example, to view 123.htm, the system creates display_page 1.htm 642. This page contains some standard HTML tags, including <HTML> and <BODY>, and some generated HTML that displays appearance preferences and metadata suffixed by the entire 123.htm. This simple blind concatenation is possible for two reasons. First, when the system creates these individual HTML files, this makes them do not contain any document global information, so that the files can be arranged endlessly. The system creates them so that their modes refer to the appropriate subdirectories, and there is no need to reprocess related file references as a result, because all entries and generated files are at the same directory level.

7A is a flowchart of an embodiment of a method for creating a context summary based on a user's context database. The use of metadata provides an information context to enhance targeted advertising and search. The method may be embodied as computer instructions executing on a processor and may be stored on a computer usable medium.

In one aspect of the invention, the context summerizer 332 distills "usable" pieces of information out of the context database and creates a concise summary of interest to the user at any given moment. For this embodiment, the database summerizer can analyze the relationships defined in the context database and the metadata associated with each item, and generate a weighted, keyword-based summary of interest to the user. This summarization method effectively provides a multidimensional synopsis of interest to the user in a format that can be effectively analyzed by a computer.

In the embodiment of FIG. 7A, metadata, in particular nominal keywords associated with each item, are analyzed. Each item is represented by an entry in the object table 380 and a keyword is represented by an entry in the tag table 392. Each item entry has zero or more keywords as defined in tag relationship table 392. Each item belongs to one or more collections as defined in relationship table 382. Each collection has one or more parent collections as defined in relationship table 380.

In the illustrated embodiment, the context summerizer creates a context summary based on two types of input of information or seeds, the first being a selection of an item or a collection, which can be represented by a user click, and the second being One or more words entered by use. In the example shown, if the seed is of the second type, the method includes the actions used for the selection of the item or collection as described in FIG. 7A.

In the embodiment of FIG. 7A, when a user input indicating selection of an item or collection is received, the current variable level is set at the level of the item or collection selected by the user. Next, keywords related to the item or collection of the current level are obtained (704). At this level all keywords are weighted according to [100 / (level−0.05)] which prioritizes the keywords closest to the root level (706).

In this embodiment, by reading the tag relationship 384, a determination is made whether the keyword is associated with an item or collection in the current open tree (708). The current open tree is a path in the file system hierarchy from the top level collection, such as All Collections in FIG. 2A to the root level of an item or collection. For example, if a user clicks on the item "US Intellectual Property Symposium" 228, the current open tree is "US Intellectual Property Symposium" 228, PTO Collection 226, Patent Collection 224, and All Collections. (All Collections) includes a collection 223. If a keyword is associated with an item or collection in the current open tree, a context filter is used in keyword 710 to provide additional weight to this keyword. The context filter adds additional weighting to collections and items in the current activity path. This allows the present invention not only to know the user's interests, but also to know the specific interests of the user at any given moment.

After each keyword is weighted at the current level 711, the name of the item or collection 712 has the lower weight used for it 712. For example, the weight of the name is determined by [75 / (level) 2-0.05)]. If it is determined that the item or collection is currently in the open tree 714, the context filter is also used for the name 716. If there is another level up in the file hierarchy 718, the current level is set until the current level is set up to the next level up (720) and the process is repeated until there are one or more levels to traverse (722). .

Name / keyword weightings are cumulative, and if the same keyword is used to define three items in the hierarchy, their respective weightings are added together (719).

FIG. 7B illustrates an example of creating a context summary using the embodiment of FIG. 7A. The levels of the tree are assigned in increasing order of 1 for the root level 748, 2 for the parent level 746, 3 for the grandparent level 744, and 740 up to level 5. The example uses keyword weighting defined by [100 / level-0.05]. The weight of the keyword is highest at the root level 748 and increases somewhat for each level going up in the tree. 0.05 is subtracted so that a keyword that is twice as high at level 2 746 is given a weight that is slightly lower than the wett of the keyword at level 1 748. If this factor is not present, order in this case (ordering) is arbitrarily because the two keywords added together are weighted at (100/2) + (100/2) = 100, which is exactly the same as a single keyword at level 1: 100/1 = 100. Using a factor of 0.05, these weights change to 10 / 0.95 = 105.3 for the root level keyword and (100 / 1.95) + (100 / 1.95) = 102.6 for the keyword that is twice as high at level 2.

As shown in FIG. 7A, a title is given with less weight and in this example the title weight is defined as [75 / (level ² ) -0.05]. The weight of the title is highest at the root level 748, but lower than keywords found at the same level. Also, the title's weight decreases rapidly as it moves up in the tree. Note that the 0.05 factor exists as above for the same reason.

In this example, the context filter is inserted as an additional multiplication factor. If the item or collection being estimated is in the current open tree as indicated by an asterisk, the weight will be multiplied by 2.5. This argument was chosen so that the keyword / title from an item or collection at level 4 in the current open tree is further weighted from another item or collection at level 2 where the current open tree is not.

Note that the words 750, 752, 754, 756, 760 in the current open tree are weighted larger than the words on the non-current path to the fifth level. This means that arguments of 2.5 take precedence over some keywords over others. Note that "tricycles" 760 are weighted lower than "truck" 758 even though they are currently in an open tree path.

The weights of the word / phrase are added if they already exist in the current path as shown by the weight of " tires " 752 in the above example.

8A is a flowchart of an embodiment of a method of selecting a target advertisement using a context summary in accordance with the present invention. The context summary and category are received 802 for the item or collection currently selected by the user. In one embodiment of the context summerizer 332, the category includes a prioritized list of context summary keywords. One example of how keywords can be prioritized is to use the weighted method described in FIGS. 7A and 7B.

In this embodiment, the context summary keyword with the highest priority is selected as the current keyword (804). The current keyword is compared 806 with keywords related to promotions. If a match 808 occurs, the HTML document generates 816 inserting some of the promotions of matching keywords for display to the user, such as in the Advertising Panel 206 of FIGS. 2A and 2B.

However, if no match occurs, it is determined that there are no more keywords (810). If so, the next highest priority context summary keyword is selected as the current keyword (808), and the comparison (810) checks and keywords on the matches (808) until a match is found or no more keywords exist. Repeated inspection of the number of fields 806.

If there is no match and no more keywords exist, the promotions are selected to have the same category as the selected item or collection (814), and an HTML document is generated (816).

An example of such a generated document is the advertisement 970 for the lowest available fare of FIG. 9C that appears when a user clicks on African (972) in the Travel collection (974).

Another aspect of the invention provides for "searching in context." When an item or collection is selected and a search of the Internet is requested to be performed in the context of the selected item, the context summary is used to synthesize a powerful search query. This query, which is generally much more complex than the user typically uses, is passed according to existing search portals. Since inquiries provide search portals with more information dynamically than the most manually entered inquiries, search engines typically return results containing more appropriate hits.

8B is a flowchart of an embodiment of a method of providing an enhanced search string to a user. In one embodiment, this method is performed by search enhancer 346. There are two ways a user can initiate a search. The user may click on an item or collection and request a search by clicking on the web search button 220, or the user may enter a search string, for example, in a search query bar 208.

A user input indicating a search request is received 818, and a search string 820 determines whether the enter has been entered. If not determined, the item or collection is selected, so a context summary is made 824 for the selected item or collection. If the search string has been entered, the search string is resolved to the user's search words (822).

Keywords in the user's context database 140 are searched for the occurrence of the user's search words (826). If no occurrence is found, the method is removed 827 in this embodiment, in addition, the items and collections have the user's search words 830. For each item or collection, a context summary is generated 832. The weights of keywords that appear more than once in both context surrogates are summed (834).

Created based on the results of context summarizing by the user clicking on an item or collection, entering keywords, and selecting a search summary up to the maximum number N of keywords with the highest weights as keywords (836). do.

Next, the number of keywords with weights that pass the threshold, i.e., the number not exceeding the second maximum number Q of keywords, is designated as priority search words (838). Search summary keywords are located in the search query grammar protocol of the user selected search engine 840 with grammar specific emphasis given by priority search words. An example of such a protocol is a "+" sign before the word to indicate that the word should be included in the result absolutely. The search summary keyword is then displayed to the user with priority search words marked for the user (842). An example of a list of summary keywords 963 is shown in FIG. 9B. The plus "+" sign in front of "vacation" 960 and map "966 indicates that they have been suggested to the user as priority search words. Select simply any of the words listed under "More Suggestions" by simply choosing from.

9A shows a user interface window in which the user clicked on "Elephants". The context summerizer has a weight for the name "Elephant" 926, a lower weight for the name "Masai Mara" 918, a lower weight for the "Kenya" 914, Add weighting up to “Trips” 910. A similar reduction in weighting is used for keywords associated with each collection or item that we encounter as we traverse the hierarchy upwards. Note that the metadata associated with Tanzania (920) is not affected by the weight at all in Kenya's Masai Mara (918) context, so the context summary for this click It may include:

Travel (keywords related to all collections under travel)

Africa (keywords related to Masai Mara, elephant, and collection name)

Masai Mara (keywords for names of elephants and collections)

Kenya (keywords for the names of Masai Mara, elephants, and collections)

Trips (keywords for Africa, Masai Mara, and collection name)

To highlight the impact of the context in determining the final weighting, the example below where the "elephant" item remains in multiple collections, the circus collection somewhere in the Masai Mara collection and the "USA" collection shown here Consider.

If the user clicks on an elephant in the circus collection, the summerizer does not suggest something like "Kenya" or "Masai Mara". Rather, these words are incorporated by "circus", "training", and the like.

The foregoing description of the embodiments of the invention has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Many variations and modifications are possible in light of the above teachings. It is intended that the scope of the invention be limited not by this detailed description, but rather by the claims appended hereto.

Claims (23)

A computer implemented system for capturing and managing granular information from digital sources, A user interface for communicating with digital sources, Capture control for capturing items of individual information from digital sources, An item storage buffer accessible by the capture control to hold the item; And an item access system adapted to provide access to the item and to store the item. The computer implemented system of claim 1 wherein the item access system is a file system. 3. The computer implemented system of claim 2, wherein the file system has a uniform file formatter for converting each item captured in a uniform file format. 2. The computer implemented system of claim 1, further comprising a metadata generator for analyzing the item generation metadata associated with each item. The computer implemented system of claim 4, wherein the metadata includes a name, keywords, category, source, and source description. 3. The computer implemented system of claim 2, wherein the file system saves external files that are modulated by an item in local files associated with the item. In the system for providing a context database, Item access system, With the metadata generator, And a database manager in communication with the item access system and the metadata generator, the database manager accessing an organized storage area according to a database for storing items and the metadata associated with each item. system. 8. The system of claim 7, wherein the item access system is a file system having a file hierarchy, wherein each item in the database is stored in a collection. 10. The system of claim 8, wherein each item has a uniform file format. 8. The system of claim 7, wherein the database is a SQL database including tables defining relationships between related items and their metadata. The system of claim 10, wherein the relationships are defined by tables. In a system for providing a context summary of a context database, A context database containing relationships between items and their associated metadata; And a context summarizer for analyzing the relationships of the selected item adapted to create a context summary. 13. The system of claim 12 wherein the items are stored as files in a hierarchical file system. 14. The context summary providing system of claim 13 wherein the context summerizer assigns weights to keywords that prioritize keywords of items in a file hierarchy closest to the selected item. 15. The system of claim 14, wherein a context filer is used for their keywords in the current path of the selected item at the top level of the file hierarchy. 13. The context summary providing system of claim 12, wherein the context summerizer assigns weights to names that prioritize names of items in the file hierarchy closest to the selected item. 17. The system of claim 16, wherein a context filer is used for their names in the current path of the selected item at the top level of the file hierarchy. In a system for providing a search query, A context summary including priorities of keywords and names for the seed, And a search query generator that accesses the context summary adapted to select the highest priority keywords and names for the query. 19. The system of claim 18, wherein the seed is a selected item or collection. 19. The system of claim 18, wherein the seed is one or more user entered query words. 19. The system of claim 18, wherein the search query generator further comprises a syntax generator for transforming the generated query into the appropriate syntax of the selected search engine. A system for providing in-context advertising to a user, An anonymous profile analyzer having a context summary and adapted to determine the best promotion source, And a promotion interface for generating an HTML document that inserts results from the promotion source. A method for capturing and managing information from digital sources, (a) capturing an item of individual information from a digital source, (b) generating metadata associated with each item, (c) forming the item in a uniform file format, (d) storing the item as a file in a file system identified by metadata associated therewith, (e) integrating the metadata into a WebBase; (f) synchronizing the metadata to be inserted into the file of the item as well as the web base, (g) creating a website summary by analyzing the metadata associated with each item and the relationships defined in the website, (h) focusing the user's current active information by context filtering, (i) creating a context summary using the context summerizer, (j) suggesting an improved search based on the context summary, and (k) making targeting advertising to the user based on the context summary.