AU2039001A - Method for extracting digests, reformatting and automatic monitoring of structured online documents based on visual programming of document tree navigation and transformation - Google Patents

Description

WO 01/14951 PCTIUSOO/23140 METHOD FOR EXTRACTING DIGESTS, REFOMWATTING, AND AUTOMATIC MONITORING OF STRUCTURED ONLINE DOCUMENTS BASED ON VISUAL PROGRAMMING OF DOCUMENT TREE NAVIGATION AND TRANSFORMATION 5 FIELD OF THE INVENTION The present invention relates to a method for extracting digests, reformatting, and automatic monitoring of structured online documents based on visual programming of document tree navigation and transformation. More particularly, the invention relates to a system and 10 method whereby a user selects a fragment of an online document shown in a source window and copies this fragment to the target window, the system creates a sequence of commands that can reproduce this behavior when applied to the new versions of the source documents downloaded from the information source, such as web site. 15 BACKGROUND OF THE INVENTION Structured online documents, especially HTML and XML documents available on the World Wide Web (WWW) have become very important in the past few years, Such documents contain data which may be periodically updated, wherein such updating does not substantially change the format of presentation of such data. 20 Those online documents usualy are dynamically generated by the web servers and they present data stored in online databases. This data periodically changes, but since these documents are automatically generated by computers, the presentation document structure remains WO 01/14951 PCT/USOO/23140 2 substantially the same for relatively long periods of time. Additionally, even when the web page is updated manually, the presentation document structure may remain substantially the same for relatively long periods of time. s Examples of such frequently updated online documents include: stock quotes from brokerage web sites; prices of specific items from online commercial vendor sites and from online auction sites; local weather information from weather web sites; airline ticket information provided by airline or travel sites; shipment tracking information from the mail delivery companies; current news headlines from the news organizations web sites; latest press releases of 10 4 specific company issued on their web site; bank account balances for an individual or corporation from the bank web site. While all this data may be of great interest to the user, it is often accompanied by data that is unimportant or even irrelevant to a particular user. This irrelevant data unnecessarily 15 complicates comprehension and interpretation of the relevant data and often leads to the user missing important changes in the relevant data. Examples of the data that may be unimportant to the user are: 1. Stock quotes for a stock of interest to the user are often accompanied by other 20 data such as number of shares outstanding, opening and closing prices, earnings in the last quarter and so on. While the user may need to check this data once every 2 or 3 months, the user is not likely to want to see this data every time a current stock quote is sought.

WO 01/14951 PCT/USOO/23140 3 2. Fluctuating price for an item in an online store that interests user may be accompanied with advertising for other items that the user has no interest in or it may be accompanied with product photographs which user has already seen many times. 5 3. Balances of the user's bank accounts may appear in separate online documents (web pages) and be accompanied by the last 10 transactions. The user, however wants to monitor only balances of all his or her accounts in the bank so that every balance appears in a small window unaccompanied by any other information. 10 In addition to this, if the user wants to monitor important data. be or she will find it necessary to push the browser "Reload" button to obtain the latest data from the remote database. This requires considerable manual effort and can be fatiguing even when monitoring one online document. The manual effort required for monitoring several online documents simultaneously is 15 so great that it makes such monitoring very difficult, if not impossible to do on a regular basis. Summary. Online documents generated by online databases provide valuable data that a user may want to monitor. However, this essential information is often accompanied by large quantities of non-essential and even irrelevant information, or information that rarely changes and 20 does not need to be monitored. Therefore, a method is needed that allows a user to automate monitoring of essential data WO 01/14951 PCT/US00/23140 4 extracted from online documents while ignoring non-essential or irrelevant data. In the remainder of this Section we present the state of the art in the technical area of this invention and show how this invention differs from the state of the art. 5 HTML. browsers. and DOM HTML, and XML structured online documents are displayed using web browsers such as Navigator by Netscape@ Communications (httpJ//www. netscape.com) and Internet Explorer by Microsoft corporation (http://www.microsoft.com/). 10 A web browser is used in the preferred embodiment of the present invention. However, none of the browsers known to us can display a document fagmena in a separate window with no window treatments so that irrelevant information is not seen by the user 15 and this window takes small space on user's screen. Also none of the browsers known to us implement automatic refresh. The present invention augments the browser behavior and it uses the ability of the more advanced browsers to be controlled by other applications. Also the present invention uses the 20 Document Object Model (JDOM) to navigate the content of an online document represented as a tree of nodes.

WO 01/14951 PCT/US00/23140 5 Web site server-side customizations Most major web allow limited server-side customization of their content these days. Examples are MyYahoo@b (http://www.yahoo com/), MyNetscape@ (http://www.netscape com/), etc. These customizations are nothing more than accounts created for users on these web sites. 5 Users see the customized content when they login into their accounts on the web site. Web site customizations provide a limited choice of what can be customized. For example, the user usually can select a portfolio of stocks to be displayed, but he or she usually cannot select what parameters are presented for a particular stock. Also usually such 10 customization are limited to very few online data categories. For instance, user can monitor all U.S. stock using such customization, but he or she cannot monitor, say, Brazilian stock even though online stock quotes for Brazlian stock may be available online. Furthermore, creating user-customized web site content requires complicated and 15 therefore expensive programming from the web site maintainers, so this option is not practical for smaller web sites because of its price and complexity. Finally, server-customized web pages are still shown in a regular web browser window that has a lot of unnecessary window treatments and user is still required to push the "Reload" 20 button every time he wants to update. Using the present invention, the user can arbitrary customize and monitor any web page WO 01/14951 PCT/USOO/23140 6 content and select any presentation format for the customized content, and no programming is required both on web server side and on the user side. Online data oroviders 5 Several online services exist that can push certain online data such as stock quotes to the user's wired or wireless device such as pager or computer. These services compare to the present invention in the same way as server-side web site customizations, because they have the same problems: limited choice of content that can be 10 monitored, no way to arbitrarily customize presentation of such content and what parameters are included, expensive server-side programming is required. XML and XSLI Several techniques exist that transform a higher level abstract document presentation to 15 the lower level document presentation used for rendering the document. Most notable effort in this area is XSLT language (httpJ/www.w3c.org) that is used to write programs that transform XML documents (http/www3coral to HTML documents that are rendered in a web browser. These techniques do not cover the present invention because they are used to synthesize 20 lower level document presentation from the higher level document presentation but they do not change the content of the document. The present invention is primarily used to change the content of the document without changng the level of abstraction used in the document presentation.

WO 01/14951 PCT/USOO/23140 7 Related Patents U.S. Patent No. 5,530,852 to Meske, Jr., teaches how to build web sites that store news articles and serve them to users through the Internet, providing categorization and search 5 services. A typical news article is a structured document that has a title, summary (profile), and body. However, the patent No. 5.530,852 teaches processing news articles in the web server space, and not in the client space. Also the patent No. 5,530,852 teaches programming of reformatting by a highly skilled computer programmer, whie the present invention teaches creation of reformatting script by non-programmer user. 10 U.S. Patent No. 5,737,592 to Nguyen et al. teaches how to build server-side programs that receive queries from a web browser, automatically convert them to SQL queries, run these queries on a database, convert records returned by the database to HTML and send this HTML back to the requested. The present invention is different from this patent because it applies on the 15 client side and not on the server side and we are not concerned with generation of SQL queries. U.S. Patent Nos. 5,745,754 to Lagarde et al. and 5,752,246 to Rogers et al. teach how to build server-side programs that use Distributed Integration Solution servers to perform extraction of data requested by-a user from databases, and presentation of this data in HTML. These 20 teachings would be of use to a highly-skilled programmer who programs web applicatiotis in extracting and reformatting data in a database. But they are different from the present invention, because we teach how non-programmer user can create reformatting scripts on the client side.

WO 01/14951 PCT/US00/23140 S U S. Patent No. 5,774,123 to Matson teaches how to record a sequence of navigation commands performed by a user on the web browser and how to later replay these commands causing the browser to repeat the navigation session. The record-and-replay feature of this patent 5 does not teach extracting digests of online documents, nor does this patent teach extracting document digests using document trees and displaying the digests in a separate window U.S. Patent No. 5,799,304 to Miller teaches how a user agent can filter, i.e. wholly display or wholly reject, a news article based on criteria provided by the user. That is, it teaches how to 10 make search engines more intelligent by using agent technologies. This patent does not relate to extraction of document digests. U.S. Patent No. 5,890,152 to Rapaport teaches how to build a web search engine that takes into account user characteristics such as IQ, etc., all stored in a personal profile database. 15 This patent does not relate to the present invention, because we are not concerned with user characteristics at all. U.S. Patent Nos. 5,895,476 and 5,903,902 to Orr et al. are concerned with server side generation of online-documents from the specialized higher level representations of documents. 20 This is different from the present invention because the present invention applies on the client side and it does not change the transformed document's level of abstraction.

WO 01/14951 PCT/USOO/23140 9 Accordingly, it is a problem in the art to automatically monitor user-selected fragments of the online documents and to create scripts that perform such monitoring when such scripts are to be created visually by a usw without requiring user to write a program of any knd. 5 SUMMARY OF THE DWENTION From the foregoing, it is seen that it is a problem in the an to provide a device meeting the above requirements. According to the present invention, a device is provided which meets the aforementioned requirements and needs in the prior art, 10 Specifically, the device according to the present invention provides a method for extracting digests of structured online documents, and automatic monitoring of the said digests. A digest of an online document is a collection of fragments of this document which are of interest to a user. Creation of the scripts that perform tbq said digest extraction and monitoring employs visual programming of the onlin, document tree navigation and transformation. The disclosed 15 method can be applied to structured online documents such as HTML, XML, SGML documents, or to any other online document that has internal structure that can be represented by a tree. More specifically, the system according to the present invention is based on a visual programming whereby a user selects a fragment of an online document shown in the source 20 window and copies this fagment to the target window that contains the reformatted digest. The system according to the present invention generates a sequence of web site navigation commands, online document tree naviption commands, and copy commands that cause the assembly of the WO 01/14951 PCT/USOO/23140 10 reformatted digest in the target window. The user can later ask the system to replay the sequence of generated commands, thus causing automatic creation of the reformatted digest of the changed version of the online document. 5 Therefore, according to the present invention, when content of the original document changes and the script that creates the digest is run, the change is automatically propagated to the digest document. This allows implementation of simple automatic monitoring of digests of the online documents which occurs entirely in the user space, that is in the application that controls the user's browser. 10 The digest document is typically much smaller than the original document, and usually it does not contain computationally intensive and bandwidth intensive multimedia elements such as graphics, sounds, scripts, and controls. This considerably lowers the screen size, bandwidth and processing power requirements for user agents that display document digests. Therefore, 15 documents digests can be displayed by user agents that run on wireless and portable computing devices. Such devices have small screen, and their bandwidth and computational power resources are limited. The preferred embodiment of the present invention is a computer program that is called 20 WebTransformerTrm. It runs on Microsoft I Windows @ 32-bit operating systems and as of fMling date it controls the Microsoft Internet Explorer.

WO 01/14951 PCT/US00/23140 11 Vocabulary Source Document and Source Window. The source window typically contains a regular browser such as Microsoft Internet Explorer. In this window the source online document is shown. Used to navigate to the web page of interest and to select a fragment of this page to be 5 monitored. Target Document and Target Window. The target window is where the digest of the source document is displayed. The digest of the source document that user monitors is also called the target document. The target window is typically much smaller than the source window and it 10 does not have window treatments such as menu bars and scroll bars, so that it is possible to have many such window on one screen. Command - Elementary instruction to perform operation on a document tree that can be recorded. 15 Script - A recorded or otherwise created sequence of commands. How It Works The user typically performs the following actions in order to use the present invention. 20 First, the user browses documents in the source window and when seeing a document of interest selects a fragment of the document that constitutes a digest. Selection is performed by clicking the desired element of the web page. This click is translated by the browser into the WO 01/14951 PCT/USOO/23140 12 address of the node in the document tree that represents the minimal HTML element that covers the clicked area. The user can then use the arrow keys of a computer keyboard to extend, contract, or 5 move sideways the selection. Other selection mouse cUcks and keyboard keys may be used depending on the web browser. When the user finishes selecting the fragment, the user invokes the user inter'ace "Copy" command that copies contents of the selected fragment from the source window to the target 10 window. In addition to that, according to the present invention the WebTransformer creates a script that records the source document location, sequence of document tree navigation commands that leads from the tree root to the node that corresponds to the selected fragment, and the "Copy To 15 Target" command. The system can record all elements of user navigation including entering User ID and Password or filling out and submitting other online forms that cause the desired navigation. 20 Finally, according to the present invention the user can ask the WebTransformer to run the script that has been created. The user can request a one-time execution of the script or automatic periodic execution of the script according to a user-specified time table. Script execution results WO 01/14951 PCT/USOO/23140 13 in fresh (not from cache) download of the source document, navgating the source document tree to the selected tree node and copying the selected source document fragment to the target window. 5 Summary of Benefits The present invention brings the following benefits to its user: 1. User views and monitors only the fragments of online documents that are of interest to him or her, not the whole documents. 10 2. User does not have to push the "Reload" button, it is done for him or her automatically by the WebTransformer. 3. Combination of typically small size of target windows and auto-refresh feature allows to monitor many (10-50) online documents simultaneously without applying any manual effort. is 4. Since the document digest is small and it typically does not contain large pictures or embedded programs (such as JavaScript, Java, ActiveX programs), the document digests download and execute much faster than the original documents. 20 5. Since document digests are small in size, and since they require less bandwidth and less computational power to display than the original documents, the document digests can be successfully displayed on small-screen user agents that have bandwidth and computational power WO 01/14951 PCT/USOO/23140 14 limitations, specifically on user agents that run on wireless devices such as cellular phones, pagers, wireless personal digital assistants (PDA), and so on. These devices' primary limitation is screen size, so they would geatly benefit from the present invention. 5 Other objects and advantages of the present invention will be more readily apparent from the following detailed description when read in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 schematically shows two source documents, each shown in a source window, and a 10 target document shown in a target window. Fig. 2 shows a concrete example of source document from the financial web site contained in source window and the document digest of this document shown in a target window. Fig. 3 shows a concrete example of source document obtained from a shipping company and digest of this document monitored in a target window. It also shows several other target 15 windows that monitor other source web pages with their source windows hidden. Fig. 4 shows a partial source document tree for the source document shown in Fig. 2. Fig. 5 shows a WebTransformer script that extracts document digest from the source window and shows it in the target window in Fig. 2. Fig. 6 shows a block diagram for client-server WebTransformer setup. 20 Fig. 7 shows a block diagram of communicating devices for use in a wireless device application according to the present invention.

WO 01/14951 PCTIUSOO/23140 15 DETAILED DESCRIPTION OF THE INVENTION Windows 5 In the preferred embodiment a user typically observes two windows per instance of the WebTransformer script: 1. Source Document Window. This window contains the source online document that is displayed using a regular web browser such as Microsoft Internet Explorer. This window is used to navigate to the online document that will be monitored and to select a fragment of the 10 online document to be monitored. 2. Target Document Window. This window is where the digest of the source document appears. This window is usually smaller than the source window and it typically has no window treatments such as menu bars, control box, or scroll bars. 15 When a WebTransformer script is recorded, both source and target window are displayed. When the recorded script is replayed, user has an option of displaying both source and target window or only the target window. Typically user does not display the source window at the script replay time. 20 If target document is assembled from several source documents, then several source windows may be displayed. However, each WebTransformer script typically has only one target window associated with it.

WO 01/14951 PCT/USOO/23140 16 The goal of this design is to keep target windows as small as possible so that several such windows monitoring different documents can be placed on the screen without overlapping each other. Target windows also can be placed on the system tray of the Mcrosoft Windows@ 5 operating systems. Figure 1 schematically shows two source documents in source windows and one target document in the target window. Source document I is displayed in the source window 10. Source document 2 is displayed in the source window 20. Target document is displayed in the 10 target window 30. Figure 2 contains the actual screen shot of the working WebTransformer. It shows the source window 10 on the left that contains the source online HTNL document from the web site at "http:/www/quicken.com/' that contains a detailed stock quote for CyberCash@ Inc. Note that 15 the "Last Trade" digits "12 3/4" (30) are highlighted to show that these digits constitute the document fragment selected by the user. The small window 20 on the right is the target window that shows the target online document that contains the same digits "12 3/4" (40) that constitute the target document fragment 20 that was copied from the source document fragment 30. The target window title contains the name of the WebTransforner script that created the target document and the time when the script was run last time.

WO 01/14951 PCTUSOO/23140 17 Figure 3 shows the web page (online document) 10, in this case depicting a FedExV Corp. web page that is used to track air shipments. A user selected web page fragment 30 that contains the latest event that happened to the user's shipment. This fragment is copied to the target 5 window 20 where it is shown as the document fragment 40. Also shown in Figure 3 are unrelated WebTransformer target windows 50, 60, and 70 that track other web sites. Specifically, window 50 tracks stock quote taken from a financial services web site, window 60 tracks a particular lot price from the online auction, and window 70 tracks 10 weather in New Jersey from a weather web site. The source windows that correspond to these target windows are hidden on instruction from user. Source Document Tree and DOM We use tree representation of the source online document in creating the transformation 15 script according to the present invention. In the document tree each logical unit of the document such as paragraph, table, heading, emphasis is represented by a node. Node A is a child on node B if and only if the document fragment represented by node A is directly embedded into document fragment represented by node B. 20 The most popular implementation of the online document tree model for HTMfL and CML online documents is Document Object Model (DOM) (see htt:v/ww w3c org/ for details). Document Object Model is implemented in modern browsers such as Microsoft Intemet Explorer WO 01/14951 PCT/USOO/23140 18 ver 5 or Netscape Navigator ver 5. The preferred embodiment of this invention uses DOM as a source document tree model Other embodiments of this invention can use different tree models for representing the source document. 5 Figure 4 shows partial document tree for the source document 10 from Figure 2 (complete tree is too big to show it on one page) The root of the tree contains BODY element 10 that represents body of the document. The B (for bold) node 20 represents HTML element B that contains the user-selected document fragment 30 on Figure 2. The path consisting from tree nodes 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, and 41 leads from the root of the tree to the tree 10 node 20. Creating the Script A script that performs online document transformation according to this invention (also called WebTransformer Script, or WTS) is created in the following manner. i5 A source document is displayed in the first window 10 of Figure 1. The first window 10 is herein referred to as a source window 10. Transformed (target) document is displayed in a second window 30. The second window 30 is herein referred to as a target window 30. 20 A user can select a source document fragment by clicking the desired fragment using computer pointing device such as a mouse. Selected source document fragment is highlighted. Then, using keys of a computer keyboard, user can expand or contract the selected fragment.

WO 01/14951 PCTJUSOOI23140 19 In Figure 1, a fragment 15 is shown as being selected. Once the fragment 15 is selected, the user can copy the fragment 15 to the target window 30 by selecting "Copy" user interface command from the graphical menu of commands, and a 5 copied fragment then appears in the target window 30 as a target fragment 31. The user can then proceed, for example, to another online document 20, select a fragment 25 therein and copy it to another target location 32 in the target window 30. The script that downloads the source document and transforms its fragment into the 10 fragment in the target document is created according to the following rules: 1. Add to the script the "Go To URL" command that causes the browser in the source window to navigate to the source document. The location of the source document includes URL address. The location information can also include additional data that needs to be passed to the 15 web server to cause displaying of the page selected by user. such as post data and headers. The command 10 from the sample WebTransformer script shown at Figure 5 causes browser to navigate to address "http://ww.quicken.com/investments/quotesOsymbol=cych" This sample script transforms the source document 10 at Figure 2 to the target document 40. 20 2. Add to the script a sequence of "Go To Child" commands that take us from the downloaded document tree root to the document tree node that represents document fragment WO 01/14951 PCT/USOO/23140 20 selected by user for monitoring. Creation of the command sequence starts with finding a tree node that corresponds to the document fragment selected by the user. WebTransformer asks DOM implementation to compute 5 the minimal HTML element that covers the selection made by the user in document. Single mouse click is treated as a selection of zero width. Then we use parent links to walk up from the selected node to the root node. While walking, we record the indices of nodes in their parents, so that the recorded path can be walked 10 again from the root, when the document is reloaded. For instance, the commands 20, 21, 22, 23, ..., 30, 31 on Figure S walk the tree node path from the root node 10 on Figure 4 to the user-selected node 20 and on the way they pass tree nodes 31, 32, 33, ..., 39, 40, and 41. 15 3. Add to the script the "Copy To Target" command. Creating the script in the case of multiple source pages requires "Copy To Target" command to be qualified by the target ID at the target document. 20 For instance, in Figure 5 "Copy To Target" command 40 finishes the script by copying the user-selected source document fragment to the target document.

WO 01/14951 PCT/USOO/23140 21 The formal algorithm for the script creation is as follows: Input: tree node selected-element that is a part of the source document tree. 5 Output: the script object that is a list of commands. 0. Create empty script object. 1. Add "Copy To Target" command to the script object. 2. Set variable e that refers to the current tree node to selected-element. 3. Do until e is not NULL 10 3a. If e.tag is equal to "BODY" or e has no parent then Exit this loop 3b. Create "Go To Child" command object. 3c. Node p - e.parent 3e. Compute integer ix which is equal to index of node e in the node p. Index of the first child is 0, index of the second child is 1, and so on. 15 3f. Store ix in the command. 3g. Add command before the frst command at the script. 3x. EndDo 4. Add "Go To URL" command that navigates browser to the user-selected source page before the first command at the script. 20 Recorded script can be saved in a computer ile and later loaded from that file.

WO 01/14951 PCT/USOO/23140 22 Running the Script The user can instruct WebTransformer according to the present invention to run the created script or alternatively to run a script loaded from file. The WebTransformer according to S the present invention then executes the sequence of commands contained in the script, thus causing the source documents) to be downloaded from the Internet, and fragment(s) of these documents to be selected and copied to the target window. All this happens automatically according to the recorded script. 10 The user can either run the script once or instruct the WebTransformer according to the present invention to run the script automatically according to a time table set by the user (for instance, every 5 minutes). The script can be run on the same desktop computer where it was created or the script can be transferred to another computer (for example, by downloading, uploading or e-mailing it) and run on another computer. The other computer may be another 15 computer belonging to the user or can be a server computer which can run this script on a request from a client. Why the Tree? Every time we reload the source document, there is no guarantee that it will be the same 20 as the previously loaded document or that it will even be close to the previously loaded document. Many things can change even in the relatively stable documents generated from online databases: (1) Advertising banners that appear on most web pages change every time the page is loaded, and WO 01/14951 PCT/USOO/23140 23 they may have complicated internal structure that is different for every ad that is displayed, (2) Certain non-advertising items may substantially change too. For example, on Figure 2 there is a list of "Recent Headlines" Number of elements in this list and composition of this list may substantially change every few hours as new headlines for the company appear and old headlines 5 are removed. Also the list of available site features ("Chart", "Intraday Chart", "News", "Evaluator" and so on) changes approximately once every month as the site implements new features and removes old features. So to be able to find the user-selected fragment of the changed source online document 10 we need to rely on a document model such that an algorithm of getting to the user-selected fragment will be the least affected by changes in the other parts of the document. The Document Tree is the document model that was selected for use in the present invention, because it provides good degree of independence of the transformation script from the document changes. 15 Tree nodes and their children that are not on the path from the root to the user-selected node may change and their change will not affect the path to the user-selected element, so the script that locates this element will still work. For example, on Figure 4 nodes 51 and 52 are likely to contain the changing content, because they are related to advertising banners that are often put into IFRAMEs. But these nodes are not on the path from the root node 10 to the user-selected 20 node 20, so even if the entire content of these nodes changes, the transformation script built according to the present invention still will be able to find the user-selected element 20 in the new document tree.

WO 01/14951 PCT/US00/23140 24 However, if nodes 51 or 52 on Figure 2 are removed entirely, then the WebTransformer script will not be able to get to the user-selected node 20. Therefore repeated running of these transformation scripts in order to obtain an updated digest of the updated source online document 5 substantially relies on the assumption that the path from the root node to the user-selected fragment node will not change in the new document. This typically is the case for the frequently updated online documents, because these documents are automatically generated from the same template by a web server program which 10 uses the same template for dynamic online document generation. Client-Server WebTransformer In the present invention, as described above, displaying of the document digest occurs in the same process and on the same computer that runs the WebTransformer script and performs 15 the transformation. Under certain circumstances it becomes necessary to separate the document digest displaying function from the document digest creation function, so that these functions may be executed on different computers. Then the program that displays the document digest is called WebTrwsformer client and the program that performs the online document transformation according to the present invention is called WebTransformer server. 20 See Figure 6 for schematic drawing of the client-server setup. The WebTransformer client 10 sends a request to get the fresh document digest to the WebTransformer server 20, which in WO 01/14951 PCT/USOO/23140 25 turn sends request to download the source online document to the web site 30. When the source online document 50 is returned from the web site 30 to the WebTransformer server 20, the server performs the source document transformation and document digest creation according to the script prepared by the user and uploaded to the server and the resulting document digest 40 is 5 sent back to the requesting client. The client-server WebTransformer can be used in the following situations: 1. WebTransformer client is located on a small-screen handheld or wireless device. Wireless provider or individuals themselves setup a WebTransformer server and put their 10 WebTransformer script on it. The wireless device client connects to this server to get the document digests. This setup is described in more detail below. 2. A company sets up a firewall that does not give any access to the outside Internet to company employees but uses Internet web sites to feed only the approved information to the employees. The company sets up WebTransformer server 20 and puts on it a number of 15 WebTransformer scripts that extract and reformat the approved data from the Internet. The access to the outside Internet is closed to employees, but they can use their WebTransformer clients 30 to view the approved document digests from the WebTransformer server 20. 3. A company sets up WebTransformer server that monitors a particular web page or assortment of web pages that are of interest to the company. The documents digests extracted by 20 WebTransformer scripts are read by robotic client that converts them to text and stores them into database. This is a good way to arrange important data extraction through the web site.

WO 01/14951 PCT/USOO/23140 26 Handheld and Wireless Devices The document digest produced by a WebTransformer script is usually smaller than the original document and it usually does not contain computationally intensive and bandwidth intensive multimedia elements such as graphics, sounds, scripts, and applets. This lowers screen 5 size, bandwidth and processing power requirements for user agents that receive and display such document digests. Since handheld and wireless devices such as screen cell phones, pagers and personal digital assistants (PDAs) all have small screen and most of them also have limitations in available 10 bandwidth and processing power, it is more appropriate to use such devices for online document monitoring using the present invention than to use such devices for web browsing. A complete web browser for such devices, even if developed, is not be very practical, because most web pages are designed for large desktop screens and not for small screens used in handheld and wireless devices. Therefore viewing web page designed for the big screen will not be convenient 15 on the small screen of a handheld device, and developing a small-screen version of every web page out there is impractical. The present invention provides a way of monitoring small fragments of larger web pages on a handheld or wireless device with a small screen. A preferred scheme of using the present 20 invention to monitor the fragments of the web pages on small-screen device with limitation in available bandwidth and computational power is presented at Figure 7.

WO 01/14951 PCT/USOO/23140 27 In this scheme, a user creates scripts according to the present invention on his or her desktop computer 60 on Figure 7. The created scripts are uploaded to the central server computer 20 of the wireless provider over the user desktop to wireless provider connection 70 which typically is a dialup connection. 5 The handheld device 10 can communicate with the central wireless computer 20 over a relatively slow wireless or similar link 40. The handheld device can download a list of available WebTransfonner scripts that the user uploaded to the central computer. On instruction from the user, the handheld device 10 can ask the central computer 20 to run the transformation script and to send the digest document produced by the script to the handheld device where they are shown 10 as the document digests 11 and 12. This way communications that require potentially high bandwidth, such as downloading the source online document from the web site 30 to the central computer 20 will occur over the fast communication link 50 that typically exists between server computers, all operations related 15 to the source page downloading and transformation that potentially require higher computing power will occur on the central computer 20, and the handheld device 10 will only need to download a small digest document over the slow link 40 and it will show the smaller digest document I1 or 12 on its small screen. 20 Also, the user can ask a central server computer 20 to send to the user a target document only when it changes. This way, even less bytes have to be sent between the central computer and the wireless device.

Claims (5)

1. Several source online document fragments can be can be used to create one target document. In this case, according to the present invention, the transformation script may contain several sequences of "Go To URL" commands, "Go To Child" commands, and "Copy To Target" commands that assemble document fragments from several source documents to one target 10 document. Also in this case target window contains target placeholders that designate the locations to which a particular source documents fragment is copied to. Each target placeholder has a distinctive ID and "Copy To Target" commands refer to this ID. 15

2. The target window may contain not only target placeholders but also arbitrary "document frame" content created by the user. Such additional content may be used to mark the target placeholders or the whole target document, or to additionally format the copied source document fragments. This content is created by the user with help of target document editor. Any HTNML editor 20 can be used as a target document editor. For instance, Microsoft FrontPage can be used as a target template editor. WO 01/14951 PCT/USOO/23140 29

3. A WebTransformer script created according to the present invention can be used as a means of addressing afragment of onhne document. WebTransformer script according to the present invention can be displayed on a web site or sent by e-mail. When the user clicks the WebTransformer script displayed on a web site or as an e-mail attachment, the WebTransformer 5 is automatically invoked and it displays the online document fragment designated in the script. Monitoring of the displayed frgment starts automatically after the initial display of the fragment.

4. According to the present invention, a source document fragment to be monitored by user can be addressed not only by a sequence of "Go To Child" commands that follow the path 10 from the source document root to the user-selected tree node, but also by assigning a distinct ID to the node and by using a single "Find by ID" command that finds document tree node uniquely identified by a given ID. This approach requires cooperation from the online document maintainers, because they have to assign distinct IDs to every online document element that is likely to be monitored. They can assign such IDs, for instance by using D) attribute of HTML 4.0. 15

5. According to the present invention, the WebTransformer can be instructed by the user to automatically compare the current and the previous version of the target online document, so that if they differ, the user is notified by generating alert. Such alert may results in sending c-mail message to the user-specifed recipient or in executing a program or script prepared by the user. 20 Also if the target document after being converted to plain text can be interpreted as a number, then one can generate alerts based on whether such number satisfies user-specihed alert condition. WO 01/14951 PCT/USOO/23140 30 The invention being thus descrbed, it wiU be evident that the same may be varied in many ways. Such variations are not to be regarded as a departure from the spirit and scope of the invention and all such modifications are intended to be included within the scope of the claims.