EP1008262A2 - Procede et appareil permettant d'acceder a des boutiques en direct - Google Patents

Procede et appareil permettant d'acceder a des boutiques en direct

Info

Publication number
EP1008262A2
EP1008262A2 EP98905960A EP98905960A EP1008262A2 EP 1008262 A2 EP1008262 A2 EP 1008262A2 EP 98905960 A EP98905960 A EP 98905960A EP 98905960 A EP98905960 A EP 98905960A EP 1008262 A2 EP1008262 A2 EP 1008262A2
Authority
EP
European Patent Office
Prior art keywords
product
line
pages
information
query
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP98905960A
Other languages
German (de)
English (en)
Inventor
Robert B. Doorenbos
Oren Etzioni
Daniel S. Weld
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Washington
Original Assignee
University of Washington
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by University of Washington filed Critical University of Washington
Publication of EP1008262A2 publication Critical patent/EP1008262A2/fr
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions

Definitions

  • the field of this invention relates to information access over networks, and specifically to providing assistance in accessing on-line electronic stores by automatically retrieving product descriptions in response to a user product query.
  • a service or agent should be able to access a new or changed Internet on-line store in order to automatically learn how to retrieve relevant information from the source.
  • shopbot acts as a user's intelligent assistant by tracking available network product sources or on-line stores, knowing the relevant information and features of each particular product source, and upon user request determining which : product sources are relevant to a given query, forwarding the query to the most relevant product sources, understanding the responses returned from each source, and integrating and intelligently presenting the query results to the user.
  • the shopbots of this invention possess several advantages, including the following. First, a shopbot returns only the relevant product information to the user. On the one hand, each user query is forwarded only,to the online stores in the product domain of interest to a user.
  • the invention includes a method for efficient access to product information sources on a network comprising preferably one or more of the following steps: receiving a user query for product information; determining the product information sources or on-line stores in the correct product domain relevant to this query; retrieving a description of each information source; formatting a form by which to access these on-line stores according to the query and to the retrieved description in a manner suitable for each product information source; transmitting the form to the on-line store; receiving responses from the product information sources; for each online store, understanding and extracting the relevant data fields according to the retrieved description; and presenting to the user the relevant data from each on-line store in an intelligent manner ranked by an estimate of its interest to the user.
  • this invention includes a heuristically guided process by which a shopbot can determine automatically the description of an on-line store. The heuristics are collected into domain descriptions for each product domain to be accessed.
  • the domain description include rules defining typical attributes of products in this domain and seed knowledge to generate training examples from an on-line store
  • This process includes one or more of the following steps: searching for likely product query forms at a product information source or on-line store; for each likely query form querying the on-line store both with products not likely to be carried by the store and also with popular products likely to be carried; and selecting the form for future queries which results in the greatest query success.
  • the invention comprises a computer system and apparatus for performing one or more steps of the method of this invention.
  • the user has a presentation device attached to a network to which is also attached a plurality of product information sources or on-line stores.
  • the presentation device receives user queries and displays shopbot responses.
  • the presentation device performs one or more of the steps of the method of this invention.
  • One or more of those steps not performed on this device can advantageously be performed on network attached shopbot server computers, which respond to functional requests from the user device.
  • the user device can range from a diskless hand-held terminal, to a PC, to a work station, and so forth. 4. BRIEF DESCRIPTION OF THE DRAWINGS ,
  • Fig. 1 illustrates generally a shopbot of this invention
  • Fig. 2 illustrates an exemplary user interface of embodiments of the shopbot of Fig. 1
  • Fig. 3 illustrates exemplary functional components of the shopbot of Fig. 1;
  • Fig. 4 illustrates alternative hardware embodiments of the shopbot of Fig. 1;
  • Fig. 5 illustrates an exemplary product query page for an on-line store
  • Fig. 6 illustrates an exemplary product query response page for an on-line store
  • Fig. 7 illustrates in general the learning phase of a shopbot of Fig. 1
  • Fig. 8 illustrates in more detail the learning phase of a shopbot of Fig. 1
  • Fig. 9 illustrates in more detail the shopping phase of a shopbot of Fig. 1.
  • shopbot of this invention is presented as a method for accessing on-line stores and as a system or apparatus implemented to perform that method.
  • first an overview of the invention is presented followed, second, by a detailed discussion of individual components. 5.1. OVERVIEW OF SHOPBOT ARCHITECTURE .
  • a shopbot method or system of this invention comprises software and hardware facilities that function together in one or more network attached computers to assist a user to access product information stored in network attached servers (known herein alternatively as “on-line stores,” “stores,” or “vendors”) .
  • Fig. 1 generally illustrates the relationships of a shopbot to a user and to networked on-line stores or product information sources.
  • user 1 accesses user computer 3 through standard interface devices, such as monitor 2.
  • monitor 2 In the course of work, the user needs information from on-line stores 7, attached to the user computer through various network links, such as network links 4 and 6. Since the on-line stores are many, the user can benefit from assistance in finding needed comparative product information from relevant on-line stores.
  • shopbot 5 which maintains an awareness of available on-line stores and their characteristics, and queries them through links 6 on behalf of, or as an agent of, the user.
  • shopbot 5 can partly or wholly reside on user computer 3 or be partially or wholly distributed on the network and accessed by the user through link 4.
  • shopbot is composed of three major functional modules: user interface 36, integrator 37, and I/O manager 41.
  • the user interface module interacts with the user to receive user queries for information, and to format and present information responses received from the network attached online stores.
  • the user interface is adapted to the specific product domain being accessed.
  • the integrator module accepts a user product query from the user interface module, formats it for network transmission to each on-line store of the product domain, receives product responses from these stores, understands these responses, and passes the relevant portions of the responses back to the user interface module for display to the user.
  • the integrator module is capable of accessing a new or changed on-line store and querying it in the process of determining a store description for use during the shopping phase.
  • the I/O manager module performs hardware, operating system, and network specific interfacing for the user interface and integrator modules so that porting a shopbot to different hardware platforms, operating systems, or networks requires only limited changes in well-modularized code.
  • either or both of the user interface or the I/O manager modules may be absent.
  • the functions of these modules can be already performed by other operating system components.
  • a shopbot can provide only one or more of the facilities disclosed without providing others.
  • a shopbot can simply format queries and understand responses, with the on-line store description being externally supplied.
  • a learning shopbot can be present elsewhere on the network and can provide vendor description to other shopbots on request.
  • the functions performed by the described modules may be divided or grouped in alternative fashions among a greater or lesser number of modules.
  • the processes of this invention can be implemented in a procedural programming language, such as C, or an object oriented programming language, such as C++, on the disclosed hardware configurations .
  • a procedural programming language such as C
  • an object oriented programming language such as C++
  • the user interface module has both important functionality that is common to shopbot user interfaces, whatever the product domain to which shopbots are directed, and also has adaptations to the particular product domain of a particular shopbot.
  • the preferable common functions one such is the ability to remember a user's preferences for interacting with a shopbot.
  • Such remembered preferences include, for example, screen display format including preferred product attribute fields and preferred result sort order, the number or identity of on-line stores to query, and so forth.
  • the user interface module preferably provides one or more windows on the user's screen with several defined views of the query satisfaction process along with certain common user controls, such as screen buttons, for manipulating these windows.
  • the user interface module presents lists of the on-line stores being consulted with each source symbolically represented as, for example, a network address, an icon, or another compact screen representation. Also displayed is a count of the total number of unique product items received currently.
  • clicking on the screen representation of an on-line store opens a further window with either information about this on-line store, or a display of the responses received from it, or access to the on-line store over the network, etc.
  • a shopbot user interface module preferably implements specific designs, formatting, and fields suitable to the information domain for which it is designed.
  • a shopbot for comparison shopping in a product domain of software stores can have a particular interface presentation containing labeled fields for product name, model, hardware requirements, operating system requirements, price, and so forth.
  • a shopbot for a product domain of pop/rock CD stores can have a particular interface presentation containing labeled fields for artist, group, data, publisher, .price, and so forth.
  • Fig. 2 generally illustrates the user display from an example of such an embodiment, which is further directed to the information domain of on-line, electronic software stores.
  • the shopbot display of Fig. 2 is divided into three sections. Section 11 is a title section generally indicating that this display has results from a shopbot.
  • a shopbot preferably also has a specific input query screen.
  • Section 12 presents the list of on-line stores currently being consulted represented by their WWW addresses 15, which are selectable to provide further information or direct WWW access.
  • those sources which have already returned query results are similarly represented.
  • Section 13 presents the results received so far formatted in accordance with this particular product domain into sections for the major PC operating systems.
  • Each individual item returned, for example item 16 is formatted with product name, price, and an address for the originating on-line store.
  • information display is controlled with the window scrolling and control facilities built into the web browser.
  • This user interface is implemented as a HTML formatted page created at a shopbot server and transmitted to the web browser.
  • the functions and modules can reside on the user's local computer.
  • the shopbot sends queries, receives responses, and formats results locally.
  • the I/O manager utilizes the facilities of the local operating system for user interaction.
  • the user interface is described primarily in terms of windows and buttons, one of skill in the art will recognize that this invention is adaptable to other display paradigms that provide for display of information and input of user commands.
  • the user interface module can control the entire screen and present graphical displays without intervention of a windowing system.
  • the user interface module is preferentially implemented with an object oriented programming language supplemented with a class library providing windowing functions.
  • a preferable implementation uses the Java language together with the java.awt package. See, for example, Flanagan, 1996, Java In A Nutshell, O'Reilly & Associates, sections 5 and 19.
  • the integrator Module The integrator Module
  • Fig. 3 illustrates the preferred functional modules, data bases, and functional interrelationship both of shopbot 30 in general and of integrator module 37 in particular.
  • the integrator preferably consists of three functions: learning phase modules 39, database 40 of product domain and on-line store descriptions, and shopping phase modules 38. These components are introduced here and described in detail in the following.
  • a comparison shopping query 31 is delivered to the integrator by means of the user interface module 34.
  • the integrator calls the shopping phase modules 38 to chose the on-line stores appropriate to the product domain of the query.
  • the integrator retrieves the store descriptions for these stores from database 40.
  • each shopbot can be specialized to only one product domain, in which case all the store descriptions are retrieved from database 40. In this case, if these are only a small number of stores the store descriptions can alternately be stored in a table in memory.
  • These descriptions of the on-line store and its requirements comprise a set of strings, to be subsequently described in detail, are used by shopping phase modules 38. These modules, first, retrieve product query pages from each on-line store, second, format the user query into fields on each of the pages, and third, then submits the filled-in pages to each store in parallel.
  • shopping phase modules 38 When stores 33 return responses, shopping phase modules 38, using strings in the store description, extract data from the responses and place it into a list of data fields, called a tuple format, relevant to the particular product domain.
  • each tuple can be assigned a priority order using a method appropriate to the particular user query.
  • screen display manager of user interface 36 requests data to present to the user, perhaps in response to a more- button request, the shopping phase modules pass the tuples to user interface module 34, sorted in priority order if a priority is determined. For example, if the product domain relates to on-line software stores, then the tuples optionally contain such relevant fields as product name, manufacturer, software version number, operating system required, price, etc.
  • An exemplary priority order of the tuples can be by price, by delivery delay, or other factor at user preference.
  • the user display is controlled according to stored user preferences 35.
  • a learning phase the location of a new or changed on-line store is delivered to the integrator.
  • the product domain is externally supplied to the integrator along with the on-line store identification.
  • the integrator can call learning phase modules 39 to determine the product domain of the identified store.
  • the integrator retrieves the product domain descriptions for the domain from database 40.
  • This domain description includes heuristic rules to be subsequently described which guide the learning phase modules in automatically acquiring store descriptions.
  • the learning phase modules interact with on-line store 33 in a manner tailored by the heuristic rules from the domain description in order to determine the strings of a vendor description for this on-line store.
  • a successful vendor description has been determined, it is stored in database 40 for use by the comparison shopping phase modules.
  • the I/O manager module 41 of Fig. 3 performs hardware, operating system, and network specific interfacing for the integrator module.
  • Network interfacing includes the tasks of sending requests and receiving responses from network linked on-line electronic stores according to protocols recognized by the stores. Since a preferred application of the shopbots of this invention is to shopping on Internet, the I/O manager is responsible for implementing the relevant protocols of the WWW, including TCP/IP, HTTP, and so forth.
  • I/O manager 41 can temporarily cache pages and other data in order to improve response time.
  • Operating system interfacing includes the task of window management for the user interface module and access to the database services, if present.
  • the I/O manager is constructed from commercially available protocol stacks, windowing libraries, such as the Java.awt package, and other tools.
  • the I/O manager can be performed by other system components on the network attached computer.
  • the I/O manager is designed to be scalable to multiple machines, to not require multithreaded or reentrant code, and to be cross platform and persistent.
  • Fig. 4 generally illustrates exemplary shopbot hardware embodiments and options in view of the previous general description. It illustrates the interrelationship of user computer elements 51-56, network 57, on-line stores 58, and shopbot server computers 59-61.
  • Computer 51 is a user computer including a processor, memory, and various attached peripherals. Such peripherals include display device 52, or other device for user interaction, network attachment 54, optional hard disk storage 53, and so forth.
  • Computer 51 can be alternatively a network device without permanent storage, a PC, a work station, or more powerful computer.
  • computer 51 be a PC or a work station running one of the Windows operating systems, the Macintosh operating system, or UNIX.
  • Present in the memory of user computer 51 is, among other software, local shopbot software 55 and local system components 56.
  • the local shopbot software implements one or more of the shopbot functions.
  • the local system components can include, for example, a web browser.
  • Network 57 can be any network with a plurality of attached on-line stores 58, which can be optionally conceptually classified by type of products sold into a plurality of product domains.
  • network 57 is the public Internet or a private intranet supporting the TCP/IP suite of protocols, including such user level protocols as FTP, HTTP, and so forth.
  • the on-line stores are server computers which make their stored product information available using the protocols supported by network 57. Such information can include product type, model, and manufacturers and store price and availability.
  • a shopbot can have various embodiments.
  • all shopbot functions reside in local shopbot software 55 on user computer 51, which in this embodiment must have sufficient processing and storage capabilities.
  • one or more of the disclosed shopbot functions can be distributed on other network attached computers.
  • computer 59 is a store/domain description server for accepting requests for downloading on-line store description or product domain descriptions stored in its database.
  • This database can be stored in memory or on disk using any data management system capable of storing and retrieving compact textual descriptions.
  • Computer 60 is a learning-phase server for performing the computationally more intensive tasks of determining new on-line store descriptions and providing these description either to database. computer 60 or directly to local shopbot 55.
  • Computer 61 is a shopbot server for performing the shopping modules function by accepting user queries and returning search results, perhaps using the facilities of store/domain description server 59 or learning phase server 60.
  • local shopbot software preferably only supports the user interface, which may be performed entirely by a web browser.
  • it can further include the shopping phase modules, which make query routing requests to query server 59 and store description requests to store description server 60. Further, it can include one or both of these latter functions.
  • the various computers of a shopbot system can be provided with software for performing the methods of this invention either from computer readable media or by loading across a network.
  • This invention is adaptable to known magnetic and optic media, such as disks, tapes and CD-ROM.
  • On-line stores available according to the WWW protocol send HTML formatted documents to a user in order to announce the store, display its products, and receive orders.
  • HTML document description language see, e.g., .
  • Fig. 5 illustrates exemplary product information request form 500 adapted to an wide selection of products.
  • a customer can search for products. For example, the customer can search for products of a certain category having certain words in their description by selecting a category from field.501 and entering search words in field 501.
  • the customer submits the form to the on-line store by clicking box 503.
  • Line 504 presents "navigation" aids to help the customer access other HTML information documents from the store.
  • a search form causes server computers at the store to search a database of product information describing the on-line store and then return product information to the user, also formatted as one or more HTML documents.
  • On-line stores attempt to create a sense of distinctive identity by using a uniform look and feel for their documents.
  • a particular store advantageously describes all available products in a consistent format.
  • substantially all use vertical separation and white space, that is blank areas of a document, to facilitate customer comprehension. For example, stores start each product descriptions on a separate logical line. Fig.
  • FIG. 6 illustrates exemplary product description form resulting from a query using the words "iomega jaz.”
  • each product with those words in its description is presented on a separate line, such as lines 506.
  • Line 507 is exemplary header information
  • line 508 is exemplary trailer information.
  • On-line vendors respect such regularities because they facilitate comprehension and, thus, sales to human customers.
  • these regularities are exploited by a shopbot. Presence of a search form allows a shopbot to simply find product descriptions in a store-independent manner. Regularities in the resulting product description forms permit a shopbot to learn how to access such a store substantially independently of operator assistance.. In particular, these regularities allow the learning procedure to incorporate a strong bias, and thus require only a small number of training examples.
  • a shopbot can provide assistance at on-line stores lacking such forms and regularities. However, for such stores substantial operator assistance can be required in order to tailor a shopbot for such a store.
  • shopbots assists users in comparison shopping.
  • a customer seeks the on-line store from which to purchase a particular product that is most advantageous according to some criteria. Therefore, comparison shopping identifies a group of on-line stores that sell the particular product desired by a user, and then ranks the stores in the group based on the customer criteria, e . g. , price, speed of delivery, and so forth.
  • a comparison-shopping shopbot can help answer: "find the lowest price for the Macintosh version of Abode Photoshop.”
  • a shopbot functions according to the following method. Upon receiving such a request, it determines the relevant stores and accesses them in parallel.
  • parsing and filling out these search forms is done according to a vendor description, which comprise sets of recognition strings.
  • Tne stores return to the shopbot HTML pages describing their terms for the indicated product.
  • the shopbot parses these returned pages, again preferably, according to the vendor or on-line store descriptions strings.
  • the parsing procedure ignores any header and trailer fields and parses the remaining HTML formatting code into logical lines matching a learned product description format. Returned pages matching a failure template that the shopbot has learned indicates that the search failed at this on-line store are discarded.
  • the shopbot sorts the product information, e . g . , by ascending order of price, and generates a summary for the user.
  • the total comparison-shopping problem is solved in tt o phases.
  • a shopbot analyzes on-line stores to learn an on-line store description, for handling HTML pages from the site. This phase is more computationally expensive, but is performed in advance of actual customer comparison shopping, and needs to be done only once per store. However, if a vendor "remodels" the store with different HTML formatting, providing different search forms or different product description page formats, then this first learning phase is repeated for that vendor.
  • a second shopping phase which is less computationally expensive, the learned information is actually used by a customer for comparison shopping.
  • the shopping phase is implemented according to the previously described shopbot architecture for rapid parallel access of relevant on-line stores.
  • a shopbot's implementation and graphical user interface advantageously utilizes certain important principles.
  • the shopbot in the comparison shopping phase is fast. Because most of the computational work has been in advance of shopping in the learning phase, the shopping phase can be fast. In fact, although the most time consuming step is fetching pages over the network, since such fetching is done in parallel across all vendors, a shopbot is faster than an expert human.
  • a shopbot provides its user with continual feedback informing the user which vendors are being contacted and what prices have been found so far and permitting user interrupts at any time.
  • the shopbot provides the user with enough context around any information it extracts so that the user can verify its conclusion or investigate manually. For the user's convenience, ShopBot indicates the store's home page, the search form it used, and each full product description found. In summary, Table 1 outlines the input and output to a comparison shopping task. TABLE 1 - COMPARISON SHOPPING
  • a description on the particular on-line shopping domain including information about product attributes, e . g. , name, manufacturer, price, and so forth, useful for discriminating between different products and between variants of the same product and typical popular products;
  • An attribute A e . g . , the price, by which the user wants to compare vendors.
  • a specification of the desired product in terms of the values of selected attributes, e.g., name is Photoshop.
  • input to the learning phase is items 1 and 2.
  • a shopbot is capable of learning on-line store descriptions for accessing product information in the on-line stores.
  • Items 3 and 4 are input to the shopping phase.
  • a user inputs attributes of the products of interest in a manner consistent with the domain, and the shopbot retrieves product information from the relevant on-line stores.
  • shopbot can equally well be constructed for other general shopping tasks involving network search. Further, for tasks in which the relevant online stores present significant regularities, a learning phase can be constructed which learns the relevant on-line store descriptions.
  • This subsection describes the first learning phase of shopbot processing in which a shopbot learns the information necessary for it to assist a user in the second comparison- shopping phase which can be repeatedly performed as long as the learning process preferably proceeds with a minimum of training examples and in an unsupervised manner.
  • shopbot learning retrieve a strictly limited number of HTML documents from the store.
  • operator intervention be minimized or eliminated in order to a shopbot to be able to comparison-shop at a new or revised on-line store.
  • a shopbot can learn a vendor description for an on-line store by using heuristics which exploit the regularities typically present in an on-lines store's HTML documents. These heuristics strongly direct the learning process according to these regularities.
  • these heuristics assume that every product description is somehow vertical-space-delimited, by, e.g., starting a paragraph, a new row in a table, a new line, and so forth. Such vertical- space-delimitation is specified by HTML tags such as ⁇ p>, ⁇ tr>, ⁇ li>, or ⁇ br>. Accordingly, after removing header and trailer information, the heuristics divide remaining HTML code of each page into "logical lines" representing groups of vertical-space-delimited text. Second, the heuristics assume that every product is described in the same format.
  • each "logical line” that is found is abstracted into a "line description” by removing the arguments from HTML tags and replacing all occurrences of intervening text with the variable "text."
  • the most successful such line description is used to describe that on-line store's product description pages. In fact this last regularity is expected since most on-line stores retrieve product information from a relational database with a program to create a custom information page in a simple format.
  • the learning process is product domain independent. All domain dependence is input to the learning process as data contained in a domain description. Therefore, in order to shop in a new product domain whose online stores share described regularities, the learning modules merely need to fetch the appropriate domain description from storage.
  • a domain description contains three categories of information: a description of the product attributes, heuristics for understanding on-line store pages, and seed knowledge to bootstrap learning.
  • Product attributes are categories relevant for describing products in this domain. For example, for a computer software domain, product attributes can include product name, manufacturer, price, hardware requirements, operating system requirements, and so forth.
  • Heuristics for understanding on-line store pages recognize the terminology used in the search. These heuristics are preferably of the form of rules whose antecedents match typical words used to describe input fields on query forms and whose consequents specify which product query attributes to fill into the input fields. For example, turning to Fig.
  • sample seed knowledge is used as test queries to begin learning of on-line store product description pages.
  • This knowledge includes products almost certainly not in the domain in order to learn to recognize search failure pages. It also includes common products likely to be present in many stores in order to learn successful product description pages.
  • for computer software seed knowledge can include the products "Microsoft Encarta,” “Abode Photoshop”.
  • this detailed description of the learning phase modules of this invention is directed toward their application toward learning descriptions of on-line stores, the methods of these modules are not so limited.
  • On-line stores typically provide a search form that a user can fill in with, e . g . , the name and manufacturer of a desired product and then submit, e . g . , by clicking on a
  • the on-line store responds by returning a page containing product information in a consistent HTML format for user scrutiny.
  • a shopbot accesses the page containing the search form, properly fills in the accessed search form, and then extracts relevant product information from the returned page(s).
  • the shopbot learning problem is comprised of three sub-problems.
  • a first sub-problem is that of finding the correct product search form for that on-line store.
  • a second sub- problem is to determine how to fill in the correct product search form, that is what product attributes, e.g., product name and manufacturer, to enter into which fields in the search form.
  • a third sub-problem is to learn how to extract the product information from the information pages returned from the search.
  • Fig. 7 illustrates the general process for a learning- phase shopbot. Starting with URL 551 for the home page of a particular on-line store, at step 552 the learner searches at that store for candidate HTML product search forms. It thereby determines limited set 553 of candidate search forms, F j i for further testing.
  • the learner tests each form Fi to compute an estimate E i for how successful the comparison-shopping phase would be if form F i were chosen by the learner.
  • the learner determines attribute mappings directing how to fill in the fields of the form, and then makes several "test queries," using the form to search both for several popular products of the shopping domain and also for products certain not to be in the shopping domain.
  • the results of these test queries provide, first, training examples from which the learner determines the format of product descriptions in the pages resulting from form F if including the header, trailer, and item format strings. Second, test queries provide examples of search failure pages from which the learner determines the failure string. Third, the results are also used to compute E ⁇ .
  • the learner's final output 557 consists of a vendor description: the chosen form's URL, the failure string, the attribute mappings, the header and trailer strings, and the item format.
  • the first step is to find candidate search forms at the store's web site.
  • Forms are parts of HTML formatted web pages starting with a " ⁇ form>" tag and ending with a " ⁇ /form>” tag.
  • a limit currently preferably between 25 and 100 and most preferably 50, on the number of pages a learner fetches while trying to find candidate forms.
  • the search procedure is preferably more selective. For example, if a site has 500 pages, only one of which has the right form on it, but a learner looks at only 50 of them, then a random search would have only a 10% chance of success. Therefore the learner incorporates heuristic techniques designed to increase its chances of finding the right page despite a limited search.
  • the learner prioritizes fetching of pages according to a priority scoring function. Lower scores are better, i.e., considered more likely to be the page containing the right search form.
  • a priority queue of URLs of pages to be fetched is maintained. Initially, this queue contains just the URL of the store's home page with score 0. The learner repeatedly removes the highest priority, or lowest scored, URL from the queue, fetches that page, and adds to the queue the URL's from any links contained in that page. As soon as 50 pages have been fetched, the search stops, and the learner searches for forms on the 50 pages retrieved.
  • Forbidden domains are URLs that many WWW sites are linked to, e.g., netscape.com, microsoft.com, lycos,, com, altavista.digital.com, and so forth, that are known not to be part of any on-line store's WWW site.
  • the following pseudocode illustrates the process for finding a set of candidate forms.
  • the scoring function is designed to give priority to those URLs considered more likely to contain or to lead to the desired search form.
  • Five heuristic rules are used to compute the score.
  • First, the score for a given link is always higher than the score from the page containing it.
  • the score is always computed by adding a positive number to a previous score. This is because the farther away from the store's home page, i.e., the more links followed from the home page, the less likely is the search form. On-line stores usually put the search form within 2 or 3 clicks from their home page.
  • Second, a link that contains the word "search" or "find” is likely to lead to the search form page, so it gets a relatively good score.
  • a link that points to a "text-only mode" section of the store's web site gets a good score, since a shopbot is more likely to understand text pages.
  • the links near the beginning and end of the page are more likely to lead to a search form than links in the middle of the page. This is because many pages contain long lists of links to different sections of the store; a link to a general search form page is unlikely to occur in the middle of such a list.
  • a "position- component" is added to the score.
  • the position-component is 0 for the first and last link on the page; 1 for the second and next-to-last link on the page; 2 for the third and third- from-last link on the page; and so on.
  • Fig. 8 illustrates in greater detail the process of step 554 of Fig. 7.
  • a learner selects from the candidate forms those likely to be product search forms.
  • a learner uses rules from the domain description to build an attribute mapping, which guides the learner is filling in form fields appropriately.
  • a learner queries the on- line store with dummy products in order to determine a failure format string, and at step 607, it queries the store with popular products in order to determine the format of product information in a successful product page. If any of these step fail, a learner abandons processing the current candidate form and starts with the next available form, if any.
  • a learner determines how to fill in that candidate form, that is what product attributes, e.g., product name, manufacturer, and so forth, to enter into each of the fields in the search form. To determine this, a learner first examines the HTML text of the form to extract the various type-in input fields, which are specified by HTML ⁇ input> tags, and locates the corresponding prompts, i.e., the text immediately preceding the input field which normally prompts the user concerning what data should be entered in that field.
  • a learner checks for the presence of several conditions which indicate that the form is almost certainly not a product search form. In this case, the form is discarded at step 609 without an attempt to fill it out.
  • Four such preferable conditions are described here.
  • a first such condition is that submitting the form would require accessing a forbidden domain.
  • a second such condition is that one of the type-in input fields is of type PASSWORD or TEXTAREA, as specified in attributes of the HTML tags. Such kinds of input fields are rarely found in search forms.
  • a third such condition is that the form contains no input fields of type TEXT, i.e., no ordinary type-in input fields, at all.
  • a fourth such condition is that one of the field's prompts contains such words as "mastercard,” “email,” “e- mail,” “phone,” “telephone,” and so forth.
  • the form is probably a user registration form or an order form, and not a search form.
  • Other such conditions can used as appropriate in various product domains or as on-line stores evolve and change.
  • a learner proceeds to step 603 in order to determine which product attributes, e.g., name, manufacturer, and so forth, are to be entered into each type-in input field. This determination is made differently for different product domains, i.e., for computer software stores there is one set of attributes to choose from, such as product name, version, hardware platform, operating system, and so forth, while for music CD's there are different attributes, such as artist, album title, and so forth. In general, the determination is made using a product-domain dependent set of rules that test the prompt of a field and decide what product attributes to enter into that field. An exemplary rule might test the field's prompt for the word "name,” and if found, fill in that field with the product name.
  • product attributes e.g., name, manufacturer, and so forth
  • a further exemplary rule for a product domain without part numbers might test the field's prompt for the words "part number,” and if found, leave that field blank, since the learner knows nothing about part numbers.
  • the domain description must provide a leaner with a list of such rules.
  • a learner system sequences through this list of rules and applies the first rule whose test matches the field's prompt. If a rule applies, an attribute mapping pair is added to the already found mapping pairs. The pair comprises the string in the field prompt that was matched by the rule paired with the name of the product attribute that the rule indicates should be entered in this fill-in-field. If none applies, the field is not filled in with anything.
  • Output 604 from step 603 includes the URL of the candidate form together with the constructed attribute mapping.
  • a shopbot in a learning phase determines how to extract information, in other words to parse, result pages returned from the on-line store after this filled-in form has been submitted.
  • the learner relies on several typical regularities of the product information pages.
  • the result pages typicaily are of two types: a "failure” type, when nothing in the store's database matched the query parameters; and a "success” type, when one or more items matched the query parameters.
  • success pages typically consist of a header, a body, and a tailer, where the header and tailer are consistent across pages from different product searches, and where the body contains all the desired product information, along with possibly irrelevant information as well.
  • product descriptions typically have the same unique format, not possessed by anything else in the body of the page.. Using these regularities, in order to parse a result page a learner solves three additional sub-problems: first, learning the generalized failure template; second, learning to remove irrelevant header and tailer information; and third, learning product description formats.
  • a shopbot first determines a general failure template for a form by querying the form with several "dummy" products almost certainly not in the database, such as the product named “qrsabcdummynosuchprod” from company "MadeUpManufacturerName .
  • the result page for one dummy product e.g., "qrsabcdummynosuchprod”
  • each occurrence of that dummy product name in the result page is replaced with a special placeholder, i.e., replace each occurrence of the string "qrsabcdummynosuchprod” in the result page with the string "***DUMMY-NAME***”
  • replace each occurrence of "MadeUpManufacturerName" with "***DUMMY-MANUFACTURER***” This procedure is repeated for several more such dummy products with different names. If the result pages, after replacement of strings, are the same in every case, then the learner records this page as the failure format string for this form. If the different result pages for different dummy queries are not the same, then the learner abandons further processing with this candidate form at step 609 and goes on to the next candidate form found in set 553 of Fig. 7.
  • Output 606 from this step includes the URL of the form, attribute mappings for the form, and the failure format string by which search failures can be recognized.
  • the following pseudocode illustrates the procedure for querying with dummy products.
  • a shopbot learner queries the form with several popular products from the domain as provided in the domain description, e.g., the current best-selling products in this domain. It compares each result page for these products against the failure format learned above; any page that matches the failure format is assumed to represent a failed search for this form and is discarded at step 609. If the majority of the test queries with the popular products are failures rather than successes, the learner determines that this is not the appropriate search form to use for the vendor, and it goes on to the next candidate form found in set 553.
  • the learner records generalized templates for the header and tailer of success pages, by replacing words which are product attributes with fixed standard strings, and then by finding the longest matching prefixes and suffixes substrings of the success pages obtained from the test queries.
  • the output of this process now includes the header and trailer strings by which the uninformative portions of a successful product query page can be recognized and discarded.
  • the following pseudocode illustrates this process.
  • a learner now uses the bodies of the pages from successful searches as training examples from which to determined the format of product descriptions in the result pages for this form.
  • Each such page contains one or more product descriptions, each containing information about a particular product, or version of a product, that matched the query parameters.
  • the format of these product descriptions varies widely across vendors. However, at each particular vendor, all the product descriptions usually have the same abstract format.
  • the learning phase processing searches through the possible abstract formats and picks the best one, i.e., the one it determines to be most likely to correspond to product descriptions at this site.
  • the abstract formats are described by strings of HTML tags together with keyword "text”.
  • the abstract form of a fragment of HTML is obtained by removing the arguments from HTML tags and replacing all occurrences of intervening text with the keyword "text.”
  • the bodies of success pages typically contain logical lines with a wide variety of abstract formats, only one of which corresponds to product descriptions.
  • the learner uses a heuristic ranking process to choose which format is most likely to be the one the store uses for product descriptions.
  • a preferred ranking function is the sum of the number of logical lines of that format in which some text, not just white space, was found, plus the number of logical lines of that format in which a price was found, plus the number of logical lines in which one or more of the required attributes were found. This heuristic exploits the fact that since the test queries are for popular products, on-line stores tend to stock multiple versions of each product, leading to a plurality of product descriptions on a successful page.
  • This ranking function reflects both the number of popular products that were found and the amount of information present about each one.
  • the exact details of the heuristic ranking function do not appear to be crucial, since there is typically a large disparity between the rankings of the "right” format and alternative "wrong” formats.
  • This invention is adaptable to other heuristic ranking functions that achieve similar discriminations.
  • Final output 608 of step 607 includes all the components of an on-line store description built from the current candidate form as well as the value of the ranking function. If the form is to be abandoned as described for previous steps, this ranking is set to a large negative value so that it will be ignored in the further processing. This last process of step 607 is illustrated in the following pseudocode.
  • body be the substring of page remaining after removing the prefix and suffix substrings matching header and tailer; Divide the body string into a set of logical lines, or substrings, at occurrences of vertical- space-delimiting HTML tags, such as ⁇ p>, ⁇ br>, etc; FOR EACH logical line DO BEGIN
  • a learner must chose the candidate form to use for this on-line store.
  • the learner has processed each candidate form found in set 553 according to the previously described steps illustrated in Fig. 8. For each form, it determines how to fill in the query attributes, and how to parse the result pages to find a best abstract format, that is the format that maximizes the function evaluate (format) .
  • it chooses one of those forms with the greatest value of the ranking function for shopping at this on-line store. As mentioned above, this choice is based on making an estimate Ei for each form F i of how successful the shopping would be if form Fi were chosen by the learner.
  • the Ei used is the value of the evaluate function for the winning abstract product description format. This function reflects both the number of the popular products that were found and the amount of information present about each one. Thus the form selected is the one whose best format has the greatest value of evaluate among all other candidate forms, and thus the greatest use in accessing this on-line store.
  • the learning phase Once the learning phase has chosen a form, it records a vendor description (See Table 2) for future use by the shopping phase. If the learner is unable to find any form that yields a successful search on a majority of the popular products, then shopbot abandons this vendor.
  • the learning phase runs once per merchant prior to any shopping at this merchant.
  • the learner's running time is linear in the number of vendors, the number of forms at a vendor's site, the number of "test queries," and the number of lines on the result page. The learner typically takes 5- 15 minutes per vendor.
  • Pairs of strings each pair being an attribute name and a field name, for mapping of product query attributes to fields of that form.
  • a string that matches a unique portion of a search failure page (e.g., "Product not found") .
  • the output of the first learning phase is an on-line store description, which together with the domain description is used in the second comparison-shopping phase.
  • Table 2 lists the preferred components of an on-line store description.
  • the output can be several strings used by the functions subsequently described for comparison shopping. These strings include location string, attribute mappings, failure string, header and trailer strings, and an item format string.
  • the location string contains the URL of the WWW page containing the product search form. If there are multiple forms on the search page, the location string additionally encodes which form to use for product searching.
  • the attribute mappings are pairs of strings of the form ("attribute-name", "fieldname") .
  • the "attribute names” are names of the attributes defined in the domain description, e.g., "title”, “manufacturer”, “artist” for a CD store domain.
  • the "field names” are labels for the fields on the search form to be filled in with values for the corresponding "attribute name.”
  • HTML requires a label for each fill-in input field.
  • the failure format is a single string matching a distinctive portion of the search failure page returned from the on-line store if a product search fails.
  • the header and tailer strings match the header and trailer formatting portions of a successful product search page returned from the on-line store.
  • the item format is a single string which matches an abstract version of each "logical line" of product information returned on a successful product search page.
  • An abstract version of a logical line has all text, that is everything other than HTML keywords, replaced by the string "text.”
  • This invention is adaptable to other formats for vendor descriptions. For example, it is adaptable to a language based on regular expressions for parsing HTML formatted documents.
  • the vendor and domain descriptions are used to assist a user in comparison-shopping at an on-line electronic store on the Internet.
  • this phase proceeds generally as illustrated in Fig. 9.
  • a user inputs a product information request to a shopbot, which can be, for example, purchase request 655 to find the cheapest price for a product.
  • Shopping phase modules 651 proceed generally according to the process illustrated in Fig. 9.
  • Domain description 653 for the product query is retrieved from the store/domain description database. The user request is represented as the values of the product attributes appropriate to this domain as identified in the domain description.
  • shopbot 651 retrieves the vendor descriptions for relevant on-line stores having products in this domain. Guided by vendor descriptions 654, for each vendor, shopbot 651 access that on-line stores 's product search form, fills in the form with the query attributes, and submits filled-in form 656 to the on-line store. The store returns product information result pages 657 from which shopbot 651 parses responsive product information. These results are presented to the user by user interface 650 as, for example, best buys 658 for the product of interest. In more detail shopbot 651 process according to the following pseudocode. The input to this procedure are values of the attributes supplied by the user which defines the user's current comparison-shopping request.
  • mapping denote the learned attribute-to-field mapping
  • the shopbot stores vendor descriptions, which are a repository of knowledge about on-line stores accessible on the Internet, in a description database. Since a comparison- shopper uses the shopper to comparison shop at an on-line store only after the learning phase has been completed at that store, a shopbot is able to immediately access on-line stores and quickly search for a desired product.
  • the shopbot shopper is both methodical and effective at actually finding products at a given vendor.
  • shopbot usefulness was measured by comparison with manual comparison shopping. Seven subjects were chosen who were novices at electronic shopping but who did have experience using Netscape Navigator. The subjects were divided into three groups:
  • the shopbot group completed its search task much faster than the other subjects, and generally found prices at least as low as found by the other groups.
  • the group 3 subjects limited to Netscape Navigator's search methods never found a lower price than ShopBot users. Providing the list of store URLs actually slowed the subjects down. For example, one group 2 subject failed to find a price for eXceed, and the other found a low price on an inappropriate version. These trials demonstrate a shopbot's utility for comparison shopping.
  • shopbot learning phase To assess the generality of the shopbot learning phase, an independent party not familiar with shopbot processes found ten on-line stores that sell popular software products and that have a search index at their WWW site. With the heuristics of the preferred embodiment, a shopbot was able to learn how tc comparison shop at all ten vendors. Shopbot currently shops at twelve software vendors, the aforementioned ten plus two more that were used to guide the original design. This demonstrates the generality of shopbot's architecture and learning processes and heuristics within the software domain.
  • Table 4 shows the line descriptions and heuristic rankings found during learning product description formats for two software vendors. In both cases, a shopbot picked the correct line description corresponding to product descriptions. Other vendors with different product formats have also been consistently learned.

Abstract

l'invention concerne un agent informatisé qui aide un utilisateur à accéder à des boutiques en direct reliées à un réseau. Un aspect de cette invention concerne un procédé qui permet d'acheminer intelligemment une demande d'utilisateur jusqu'à des boutiques en direct propre à la demande; d'extraire les domaines de données pertinents à partir des réponses reçues et de présenter de manière intelligente les données extraites par ordre d'intérêt estimé. Un autre aspect de cette invention concerne le système de l'invention qui met en oeuvre une ou plusieurs étapes du procédé de manière centralisée ou décentralisée sur un ou plusieurs ordinateurs reliés à un réseau. Cette invention concerne également un nouveau processus guidé de manière heuristique qui permet à l'agent d'acquérir automatiquement des informations suffisantes concernant les caractéristiques des boutiques réseautées en direct pour qu'il puisse accéder à ces mêmes boutiques et y faire ses achats.
EP98905960A 1997-01-17 1998-01-16 Procede et appareil permettant d'acceder a des boutiques en direct Withdrawn EP1008262A2 (fr)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US3562397P 1997-01-17 1997-01-17
US35623P 1997-01-17
PCT/US1998/000771 WO1998032289A2 (fr) 1997-01-17 1998-01-16 Procede et appareil permettant d'acceder a des boutiques en direct

Publications (1)

Publication Number Publication Date
EP1008262A2 true EP1008262A2 (fr) 2000-06-14

Family

ID=21883820

Family Applications (1)

Application Number Title Priority Date Filing Date
EP98905960A Withdrawn EP1008262A2 (fr) 1997-01-17 1998-01-16 Procede et appareil permettant d'acceder a des boutiques en direct

Country Status (3)

Country Link
EP (1) EP1008262A2 (fr)
JP (1) JP2001509294A (fr)
WO (1) WO1998032289A2 (fr)

Families Citing this family (54)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6226365B1 (en) * 1997-08-29 2001-05-01 Anip, Inc. Method and system for global communications network management and display of market-price information
EP0895173A3 (fr) * 1997-08-02 2003-02-12 Fujitsu Services Limited Système d'ordinateur pour rendre des services financiaux
CA2353932A1 (fr) * 1998-12-16 2000-07-06 Icu Medical, Inc. Systeme et procede permettant de rechercher par exploration et de comparer des produits
AU5034300A (en) * 1999-05-21 2000-12-12 Brodia Group, The Autonomous browsing agent
US7330826B1 (en) * 1999-07-09 2008-02-12 Perfect.Com, Inc. Method, system and business model for a buyer's auction with near perfect information using the internet
AU7880100A (en) * 1999-07-19 2001-02-05 Infospace, Inc. Intelligent mapping of field names in an electronic form with standard field names
DE19933986A1 (de) * 1999-07-20 2001-01-25 Sonja Anderle Kommunikationssystem sowie ein hierzu geeignetes Verfahren zur Durchführung des Kommunikationssystems
WO2001022255A2 (fr) * 1999-09-17 2001-03-29 Dealtime.Com Procede et dispositif de remise d'informations comparatives d'actes d'achat a l'interieur d'un reseau de communications interactives
US6381597B1 (en) * 1999-10-07 2002-04-30 U-Know Software Corporation Electronic shopping agent which is capable of operating with vendor sites which have disparate formats
US7707245B2 (en) 2000-02-22 2010-04-27 Harvey Lunenfeld Metasearching a client's request for displaying different order books on the client
US7328172B2 (en) * 2000-03-03 2008-02-05 Gxs, Inc. Provision of electronic commerce services
AU2001237574A1 (en) * 2000-03-29 2001-10-08 British Telecommunications Public Limited Company Electronic commerce
AU2000235513A1 (en) * 2000-03-31 2001-10-15 Kapow Aps Method of retrieving attributes from at least two data sources
US20030078836A2 (en) * 2000-09-25 2003-04-24 Sabre Inc. Availability based on value creation method and system
CN102708114B (zh) * 2000-09-29 2016-08-03 凯创研究有限责任公司 通过相互连接的计算机网络进行实时在线搜索处理的方法
US7555448B2 (en) 2000-09-29 2009-06-30 Victor Hsieh Online intelligent information comparison agent of multilingual electronic data sources over inter-connected computer networks
US20020103876A1 (en) * 2001-01-29 2002-08-01 Masayuki Chatani System and computer-based method for providing transformed information in response to a client search request
US20020120554A1 (en) * 2001-02-28 2002-08-29 Vega Lilly Mae Auction, imagery and retaining engine systems for services and service providers
US7966205B2 (en) * 2001-03-19 2011-06-21 Ita Software, Inc. Fare rules summarizer for travel planning
US7392237B2 (en) * 2001-04-26 2008-06-24 Siemens Medical Solutions Usa, Inc. Identifier code translation system
WO2002099692A2 (fr) * 2001-06-06 2002-12-12 Koninklijke Philips Electronics N.V. Procede de fourniture d'information selectionnee
US7536323B2 (en) 2003-03-26 2009-05-19 Victor Hsieh Online intelligent multilingual comparison-shop agents for wireless networks
JP2004343268A (ja) * 2003-05-14 2004-12-02 Nec Corp 近距離無線を用いた情報交換システム、携帯端末及びそれらに用いる情報交換方法
US7225187B2 (en) 2003-06-26 2007-05-29 Microsoft Corporation Systems and methods for performing background queries from content and activity
US7162473B2 (en) * 2003-06-26 2007-01-09 Microsoft Corporation Method and system for usage analyzer that determines user accessed sources, indexes data subsets, and associated metadata, processing implicit queries based on potential interest to users
US8090678B1 (en) 2003-07-23 2012-01-03 Shopping.Com Systems and methods for extracting information from structured documents
US20050131837A1 (en) 2003-12-15 2005-06-16 Sanctis Jeanne D. Method, system and program product for communicating e-commerce content over-the-air to mobile devices
US8370269B2 (en) 2004-06-02 2013-02-05 Overstock.Com, Inc. System and methods for electronic commerce using personal and business networks
US8719075B2 (en) 2004-09-23 2014-05-06 Overstock.Com, Inc. System, program product, and methods for online image handling
US7979340B2 (en) 2005-09-21 2011-07-12 Overstock.Com, Inc. System, program product, and methods for online image handling
US8583480B2 (en) 2007-12-21 2013-11-12 Overstock.Com, Inc. System, program product, and methods for social network advertising and incentives for same
US8768894B2 (en) * 2008-12-08 2014-07-01 Bank Of America Corporation Information management inventory
US9747622B1 (en) 2009-03-24 2017-08-29 Overstock.Com, Inc. Point-and-shoot product lister
TWI409727B (zh) * 2009-05-20 2013-09-21 Chung Shan Inst Of Science Simulation Training Machine Network Communication System and Its Operation Method
CN102859518B (zh) * 2010-02-26 2017-03-08 乐天株式会社 信息处理装置、信息处理方法
US9047642B2 (en) 2011-03-24 2015-06-02 Overstock.Com, Inc. Social choice engine
US10403391B2 (en) 2012-09-28 2019-09-03 Cerner Health Services, Inc. Automated mapping of service codes in healthcare systems
US10318635B2 (en) 2012-09-28 2019-06-11 Cerner Innovation, Inc. Automated mapping of service codes in healthcare systems
US10565315B2 (en) 2012-09-28 2020-02-18 Cerner Innovation, Inc. Automated mapping of service codes in healthcare systems
US10546262B2 (en) 2012-10-19 2020-01-28 Overstock.Com, Inc. Supply chain management system
US10949876B2 (en) 2012-10-29 2021-03-16 Overstock.Com, Inc. System and method for management of email marketing campaigns
US11023947B1 (en) 2013-03-15 2021-06-01 Overstock.Com, Inc. Generating product recommendations using a blend of collaborative and content-based data
US10810654B1 (en) 2013-05-06 2020-10-20 Overstock.Com, Inc. System and method of mapping product attributes between different schemas
US9483788B2 (en) 2013-06-25 2016-11-01 Overstock.Com, Inc. System and method for graphically building weighted search queries
US10929890B2 (en) 2013-08-15 2021-02-23 Overstock.Com, Inc. System and method of personalizing online marketing campaigns
US10872350B1 (en) 2013-12-06 2020-12-22 Overstock.Com, Inc. System and method for optimizing online marketing based upon relative advertisement placement
WO2015180775A1 (fr) * 2014-05-28 2015-12-03 GoEuro Corp. Cache intelligent pour un système informatique de recherche de voyage hébergeant un méta-moteur de recherche de voyage
US10490306B2 (en) 2015-02-20 2019-11-26 Cerner Innovation, Inc. Medical information translation system
US10534845B2 (en) 2016-05-11 2020-01-14 Overstock.Com, Inc. System and method for optimizing electronic document layouts
US10970769B2 (en) 2017-03-02 2021-04-06 Overstock.Com, Inc. Method and system for optimizing website searching with user pathing
US11436526B1 (en) * 2018-02-27 2022-09-06 NTT DATA Services, LLC Systems and methods for bot-on-bot optimization
US11514493B1 (en) 2019-03-25 2022-11-29 Overstock.Com, Inc. System and method for conversational commerce online
US11205179B1 (en) 2019-04-26 2021-12-21 Overstock.Com, Inc. System, method, and program product for recognizing and rejecting fraudulent purchase attempts in e-commerce
US11734368B1 (en) 2019-09-26 2023-08-22 Overstock.Com, Inc. System and method for creating a consistent personalized web experience across multiple platforms and channels

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4774655A (en) * 1984-10-24 1988-09-27 Telebase Systems, Inc. System for retrieving information from a plurality of remote databases having at least two different languages
AU3944793A (en) * 1992-03-31 1993-11-08 Aggregate Computing, Inc. An integrated remote execution system for a heterogenous computer network environment
US5675637A (en) * 1995-05-16 1997-10-07 Inventions, Inc. Method for automatically obtaining and presenting data from multiple data sources
US5708780A (en) * 1995-06-07 1998-01-13 Open Market, Inc. Internet server access control and monitoring systems
US5721908A (en) * 1995-06-07 1998-02-24 International Business Machines Corporation Computer network for WWW server data access over internet
US5572643A (en) * 1995-10-19 1996-11-05 Judson; David H. Web browser with dynamic display of information objects during linking
US5737619A (en) * 1995-10-19 1998-04-07 Judson; David Hugh World wide web browsing with content delivery over an idle connection and interstitial content display
US5706437A (en) * 1995-12-29 1998-01-06 Mci Communications Corporation System and method for accessing a service on a services network
US5727950A (en) * 1996-05-22 1998-03-17 Netsage Corporation Agent based instruction system and method
US5715453A (en) * 1996-05-31 1998-02-03 International Business Machines Corporation Web server mechanism for processing function calls for dynamic data queries in a web page
US5727129A (en) * 1996-06-04 1998-03-10 International Business Machines Corporation Network system for profiling and actively facilitating user activities

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See references of WO9832289A3 *

Also Published As

Publication number Publication date
WO1998032289A3 (fr) 1998-12-17
WO1998032289A2 (fr) 1998-07-23
JP2001509294A (ja) 2001-07-10

Similar Documents

Publication Publication Date Title
WO1998032289A2 (fr) Procede et appareil permettant d'acceder a des boutiques en direct
Doorenbos et al. A scalable comparison-shopping agent for the world-wide web
US7080064B2 (en) System and method for integrating on-line user ratings of businesses with search engines
US7099859B2 (en) System and method for integrating off-line ratings of businesses with search engines
US20020156685A1 (en) System and method for automating electronic commerce transactions using a virtual shopping cart
US7555448B2 (en) Online intelligent information comparison agent of multilingual electronic data sources over inter-connected computer networks
EP1389322B1 (fr) Remplissage automatique de demandes de recherches
US6256623B1 (en) Network search access construct for accessing web-based search services
US7594176B1 (en) Automated retrieval, evaluation, and presentation of context-sensitive user support
US8478792B2 (en) Systems and methods for presenting information based on publisher-selected labels
US6067552A (en) User interface system and method for browsing a hypertext database
US7702675B1 (en) Automated categorization of RSS feeds using standardized directory structures
US6546388B1 (en) Metadata search results ranking system
US6671681B1 (en) System and technique for suggesting alternate query expressions based on prior user selections and their query strings
US8849842B2 (en) Task/domain segmentation in applying feedback to command control
US7257604B1 (en) System and method for communicating information relating to a network resource
CA2361771C (fr) Systeme, procede et article permettant l'obtention de regroupement d'informations avancees grace a la technologie du web
US7031968B2 (en) Method and apparatus for providing web site preview information
US20020103789A1 (en) Interface and system for providing persistent contextual relevance for commerce activities in a networked environment
US20020042750A1 (en) System method and article of manufacture for a visual self calculating order system over the world wide web
US20030005033A1 (en) Client-server independent intermediary mechanism
US20020167539A1 (en) Portal server enabling joint exploration of the internet with an expert
EP1314098A1 (fr) Appareil et procede de production de contenus electroniques a annotations contextuelles
CA2279546A1 (fr) Explorateur de reseau intelligent utilisant un indexeur conceptuel incrementiel
WO2001024038A2 (fr) Service de courtage sur internet fondes sur des profils sanitaires individuels

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 19990727

AK Designated contracting states

Kind code of ref document: A2

Designated state(s): AT BE CH DE ES FR GB IT LI LU NL SE

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20030801