JP2002342340A - Sequential subset catalog search engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system and a method for cascading a search method in the data of segments or sets selected before. SOLUTION: In an electronic catalog search engine, the search processing is optimized and computer resources are efficiently utilized by identifying a desired item from a most advantageous provider. The search engine is provided with a search and data subset preparing mechanism. In this system, a search word is accepted from a user and the sequence of a search strategy is executed on a subset in a database. This sequence can contain a proximate search, a word count search and a fuzzy theory search. The subsets can be searched in any order as well and various search strategies can be applied to various subsets. When a search step discovers at least one of matching, this sequence is finished.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

FIELD OF THE INVENTION The present invention relates to computer software for database operations, and more particularly to a system for cascading search methods on a set of data selected from one or more electronic catalogs. And methods.

[0002]

2. Description of the Related Art Searchable electronic catalogs are commonly used in support of various electronic commerce and purchase functions. These catalogs must have a user interface for selectively retrieving data records. Engineers want to make the operation of the user interface as simple as possible. This is because the complexity of the user interface can impair sales from catalogs. What is simple is that when users with different levels of skill or training intend to access the catalog,
Especially important. In particular, the results of the search should quickly and easily direct the user to the most desirable supplier or source for the requested product.

[0003] The user interface which is easy to operate is
It should have the ability to handle almost any type of user input. In the case of an electronic catalog, if the user knows the exact part number of the desired product and enters this part number correctly into the user interface, the database search engine will use the exact search string to match , Quickly identify desired records from the database. In the more general case, the user may have only partial information about the desired product, or may enter the search string incorrectly.

Similarly, the output of a search should be easy to understand. In an era where large data accumulations are often available, there can be very large collections of catalog data to search and retrieve items. Ideally, the catalog search engine will have a mechanism for systematically searching through large electronic catalogs, and will only display the most relevant results to the user.

[0005] Excess catalog data is at least 2
Can be problematic for two reasons. (1) the desired item may be available from a number of different suppliers, thereby creating an array of output choices that unnecessarily confuses the user; When a large and fast search would find an item from a preferred supplier,
System resources are consumed to unnecessarily search the entire catalog database for desired items. Managing the output options available to the user can be particularly important in connection with companies where individual employees have the option to order their own equipment. In such a system, an administrator may want to enforce a user's hierarchy by defining a particular supplier hierarchy and displaying only the most desirable sources for the item.

[0006] Earlier systems have not adequately addressed the problem of searching for large catalog data accumulations and reporting the results in an efficient manner. No. 5,715,444 to Danish et al. Discloses a process for identifying a single item from a family of items in a database. A functional screen and search process presents the user with a guided unclassified parameter search,
Identify matching items based on user-specified criteria and priorities. Also disclosed are methods and systems suitable for the Internet environment.

[0007] Cochran et al., US Pat.
No. 9,648 and 5,206,949 disclose a method of variably displaying a search term, wherein two control inputs are used to select a plurality of words for a plurality of categories. The word at the visible position on the screen becomes a search term or qualifier for other records in the database. Search results are dynamically formed based on the selected search term. Search results can also be grouped into fixed or static lists.

[0008] More recently, Aalbersberg
U.S. Patent No. 5,946,678 discloses a user interface for document retrieval, where each query term is displayed in a different representation. In the subsequent results window, an indicator is attached to the header or title of each document, which uses the same different representation to directly indicate to the user the relative contribution of the individual query terms to each listed document. Different representations can take several forms, such as by different colors, by hatching or shading, or by displayed icons.

[0009] Efficient searching in electronic catalogs has been the focus of much effort. Previous catalog search algorithms typically use one of two search strategies. The first strategy is keyword search, which searches a database based on matching text strings.
It is for selecting a record. The second strategy is a classification search, which is based on a list of classifications to narrow and select database records.
It is for selecting a record. Each of the two search strategies has the disadvantage that it can be difficult for the user to find the desired database record.

[0010] The keyword search strategy has the disadvantage that the user must be familiar with the appropriate keywords that are likely to produce the desired data record. In addition, it is not always possible to quickly collect logically related groups of data records. If a close match is found but is not the exact match desired, it is not always possible to quickly identify all similar data records using information in the close match. Keyword search engines typically operate on close matches to identify similar items in the database, such as "more-like-
does not have this function.

[0011] The classification search strategy can take advantage of the logical grouping of data records. This search strategy is best suited for finding data that in turn logically breaks down into higher levels of detail. This search strategy works best when the data is carefully edited and constructed in the database. Finding a single related record can quickly lead to all other related records as long as the grouping logic concerns the way the data is used. Therefore, "more-l
An "ike-this" function can quickly identify all records classified in a similar manner in the database.

A disadvantage of the classification search strategy is that the user cannot always predict the proper classification of a record, and may search for the wrong category for the desired database record. Users are bound by the logical structure of the data and must learn to navigate through predefined database structures in order to locate specific data records.

[0013] Regardless of whether the search is performed with a keyword strategy or a classification strategy, the focus is on finding a particular item. In some cases, the item is available from multiple suppliers. In other cases, there may be multiple different types of items that meet the needs of the user, and may be available from multiple suppliers. In either case, it would be desirable to further refine the search method so that the most advantageous suppliers are quickly identified to the user. If not, it would also be desirable to search through a catalog of less desirable suppliers when the item had already been found, to avoid unnecessary consumption of computer processing time.

It would be further desirable to have a simple user interface, both for entering search terms and for reviewing results. On the input side, the software should allow a free-form text search, so that there is no prerequisite for format or content. Thus, even when the user enters a search string, word fragment, or other characteristic of the item being misspelled, the database record most likely to be the user's desired choice. It would be desirable to have a system that can identify On the output side, the software should only display the most advantageous sources for the item, especially when the item is available from a number of different sources.

[0015]

In many commercial situations, it would be advantageous to be able to configure search behavior for various elements. In addition to providing a simple user interface, it may be desirable to divide the database of searchable items into multiple layers. The combination of a search strategy and database partitioning will allow the identification of items from the most economical sources. This would allow the system administrator to adjust the results based on changing factors. Finally,
Such a system is efficient in computing and
Will use resources. These and other technical and business aspects of the catalog search engine are motivating factors for the invention described herein.

[0016]

SUMMARY OF THE INVENTION The present invention is a system and method for cascading search methods on preselected segments or sets of data. Each data set is paired with one or more search strategies,
The overall effect is to provide the user with the most favorable matches for the keyword search. The search strategy is
One or more of the following may be included. That is,
Exact search, stem search, soundex
ex) search and fuzzy logic search. The data sets can be preselected based on source, shipping availability, or any other business reason for selecting one supplier or source for another.

During a search, a user enters one or more search terms to identify a desired item from an electronic catalog. The search engine of the present invention uses the search method specified in its corresponding data set. The hierarchical order of the data set can be
Established based on the desire to source the item from a particular supplier or source. After the item is found, the search engine terminates the search, thereby saving computing resources from unnecessary searches in the remaining data sets.

In one aspect of the invention, the system first comprises:
The catalog (or data
Set) and then fail over to a second tier catalog (or data set) of vendor-supplied items available for short-term delivery. If both searches fail to produce acceptable results, the search engine can fail over to a custom order supplier with a longer lead time for delivery. One advantage to the user is that the first search results are often the most desirable option available.

The text search can be improved through the use of a sequential search algorithm designed to maximize the chance of identifying the desired data record. For example, several different search algorithms can be used on the most desirable data set to increase the chances of finding a suitable item in that data set. For less desirable sources, it may be advantageous to search for only exact matches of the search term.

According to the present invention, a method for selecting a data record in a catalog database includes the following steps. Inputting the search terms into a user interface, checking the search terms against a sequence of data sets using a search algorithm specified for each data set, and at least one database This is the step of terminating the sequence of the search algorithm when the record satisfies the search criteria. In some embodiments of the present invention, the algorithm can be extended by compiling a list of unique classifications from each identified record to assist the user in further refining search terms.

The present invention includes a database with a search engine. The database may consist of a collection of supplier catalogs, each data record further comprising a description composed of a category description, a manufacturer name, a manufacturer part number, a short description and parameters. data·
Each item in the record can be organized by fields.

Search algorithms that can be used in accordance with the present invention can include proximity search, string matching, stemming, fuzzy logic, and soundex matching. In some embodiments, multiple search algorithms can be performed on the data set. For example, if an exact match is found, the search stops when all exact matches have been identified and does not rely on further search algorithms. If no exact match is found, the search term is manipulated to identify strings having similar roots. Again, if no match is found, the search term is further examined according to other algorithms, such as fuzzy logic and soundex, until a match is found or the search engine reaches its logical end.

One aspect of the search strategy is that the searchable words are:
Include predefined classification terms, as well as other attributes and parameters for each catalog entry. This means that free-form text entry indicates a text string match for any classification or parameter name. This feature increases the likelihood of finding the desired data record based on a keyword search engine.

Each catalog entry can have one or more classifications associated according to type, and a list of unified classifications that can be dynamically compiled from the identified matches. Dynamic compilation refers to the process of continuously updating the list of classifications as new matches are identified in the data set. This ensures that the list is continuous and accurate, reflecting the scope of the classification of the identified matches. The list is unified in the sense that each classification is listed only once, even when the identified matches have multiple records with the same classification. A categorized list is presented to the user along with a list of matches to assist the user in further refinement of the search method.

The present invention has the unique aspect of enabling a dynamic search of a subset or block of a database by any combination of a number of search methods supported by software. The system administrator can specify which data sets or catalogs are searched first, and they are searched in that order. Each block or data set of a particular catalog can be searched by different strategies. Specific search method combinations can be assigned to users of the invention by name and password. Data suppliers also have the ability to request special priority for searches of those data types.

Those skilled in the art will appreciate that the advantages and objects of the present invention include, but are not limited to, the following. That is, providing a database search engine that can quickly and easily direct a user to a desired database record, combining the advantages of a keyword search with the advantages of a classification search, including misspelled words and word fragments. Providing an interface to handle any type of user input, increasing the efficiency of the search process by first searching in the most desirable data set, and providing a simple user・
The goal is to provide a search engine and data structure that maximizes the likelihood of finding based on the interface.

[0027] These and other features and advantages of the present invention will become apparent from the following detailed description, taken in conjunction with the accompanying drawings.

[0028]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a functional block diagram of an electronic catalog and an automatic purchase invoice system. An application server 12 is connected to interact with a database 14 residing in a computer memory storage device 16. A user of the system has a workstation 18. This is application server 1
2, via a wide area network such as the Internet. Information is extracted from the database 14 according to the command input to the web browser and displayed on the workstation 18.

The database 14 contains an electronic list of products. The database 14 is constructed in such a way that it can be subdivided and each item can be searched separately. For example, the details can be according to internal and off-premises items. Alternatively, the details may be according to a variety of suppliers, including the most preferred supplier followed by other less preferred suppliers. In yet another embodiment, the items can be arranged according to the fastest delivery date, and if the same item was available from multiple suppliers, the supplier with the fastest delivery date is presented to the user first. Become so.

A mechanism for subdividing a database provides a mechanism for identifying or grouping data records.
It can be any of the well-known mechanisms. Items need not be related to the physical storage location of the record in the database. For example, a data set can be created by having certain field values in a data record. This is used, for example, to create a data set that contains all data records from a particular vendor. Alternatively, a data set can be created by grouping certain data records into files. In yet another embodiment, software objects can be used to encapsulate some data to define sets.

In one embodiment of the present invention, software 1
0 is configured to access one of several electronic catalogs in the database 14. Database 14 is typically a collection of product information from a number of suppliers. The same item may be listed several times in aggregated catalogs from different suppliers. According to the present invention, more preferred suppliers can be searched first in the hierarchy cascading searchable data sets. Certain tiers may be provided by the system administrator based on the desirability of purchasing items from a particular supplier, or may be provided by the supplier for the privilege to be moved up the tier. Predefined based on encouragement.

There are significant advantages to having a supplier search its catalog first. This is because in some embodiments of the present invention, only the first exact match is reported to the user. Suppliers mutually searchable data
You can compete for more advantageous positions on a cascade of sets. In some embodiments of the present invention, only the first exact match is displayed to the user, and the supplier does not have to worry about head-to-head price comparisons of inventory items with other suppliers.

The type or format of each catalog in the database 14 is as follows.
It can be changed as long as it responds appropriately to text queries from zero. For example, a catalog can reside in a relational database, or can reside in an object-oriented database.

Referring again to FIG. 1, the storage device 16
Can be any disk drive, tape drive, RAM, or any known computer data storage device. The application server 12 can reside on a computer directly connected to the storage device 16, or can be connected to the storage device 16 via a network.
In each case, the application server 12 queries the database 14 and returns the results to the workstation 18
Send to

The software 10 runs on the application server 12. The software 10 follows the algorithm 200 as shown in the logic block diagram of FIG. As in block 202, the user enters a search string. In the preferred embodiment, the search string is free-form, which means that the string can be any combination of alphanumeric or search terms and no specific syntax is required. In a preferred embodiment, the search string can include the search terms in any order. For example, the search string may include the name of the item, the part number of the item, or any descriptive attributes of the item. Software 10 is designed to handle misspellings, word fragments, or any other character string that can guide a user to find a desired product in database 14.

After one or more searches have been entered,
The algorithm 200 proceeds along a logical path 204 to a first search strategy in block 206. This search strategy consists of a data set and a search method. A first search method, such as "exact string search", is applied to a first data set, such as a preferred supplier catalog. If the preferred supplier had the exact item, a match is recognized by the algorithm 200 and the logic proceeds along schematic lines 228 and 234, and the result of the match is reported to the user at block 236. The algorithm then preferably ends at block 226, but there may be additional steps associated with payment and fulfillment of the order.

If the first search method, as applied to the first data set, did not yield a match, the algorithm 200 proceeds along schematic line 208 to a second search strategy at block 210. The second search strategy has a second data set and a second search method. Generally, the second data set is the first data set
There is no requirement that it be different from the set. For example, if the first search strategy at 206 failed because there was no exact string match, it may be desirable to perform a stem search on the same data set. As such, a preferred supplier may have multiple opportunities to identify the desired item in its catalog.

Similarly, in the preferred embodiment, there is no requirement that the second search method at 210 must be different from the first search method at 206. For example, if the first preferred supplier did not make an exact string match to the entered search term, the same search method can be applied to the second preferred supplier's catalog.

The system administrator implements algorithm 200 by pre-selecting the data sets and search methods to produce the most efficient business strategy. The search strategy, or the ordering of specific data sets and search methods therein, can be updated as needed to reflect changing business conditions. For example, if a particular supplier provided a temporary incentive plan, this supplier's data set (or catalog) might be temporarily hierarchized to fill as many items as possible during the promotion. Moving to the top would be advantageous.

Continuing with FIG. 2, if the second search strategy at 210 fails to produce a match, the algorithm proceeds along 212 to a third search strategy at 214, after which a line is drawn until a match is found. Proceed along 216. The three vertical points at 218 schematically illustrate that the number of search strategies is arbitrary.

In a preferred embodiment, the system administrator extends the scope of the search strategy from "exact" string matches along a continuum of search methods, possibly including fuzzy logic strategies. In this regard, fuzzy logic algorithms are described in the art as "word w
known as "head" or other names associated with algorithms for checking for alphanumeric combinations and substitutions in search strings.
The intent of the fuzzy logic search is to identify products in the database 14 that have any descriptive similarity to the search string. For example, a user entering a search string containing a typo generally does not find an exact match in the database 14, but the fuzzy logic algorithm may select a desired product based on similarity with the search string. it can. The purpose of expanding your search strategy is to:
The purpose of the search is to fail to identify the desired item from any data set, thereby avoiding the situation where negative search results are reported to the user in logic block 222.

The software 10 corresponds to the box 30 in FIG.
It is preferable to have a single text box for the search string, such as 0. An example of the search character string in the box 300 is “bic pen red”, which represents a combination of a manufacturer name, a classification, and an attribute.
It has been found that users desire the simplicity of entering a string in a single location, regardless of whether the string represents a product name, product manufacturer, or product attributes. Alternatively, there may be separate text boxes for different types of search characters.

The software 10 executes the algorithm 200 to identify a first data set that provides a red Bic pen. If the data set is arranged such that the most preferred supplier is first, the user receives a positive search result from the preferred supplier. In this way, a system administrator can fill orders in the most efficient manner.

The software 10 can also rank the degree of similarity between the search string and each matched database record. For example, records that have a greater number of alphanumeric matches for a search term are ranked higher than records that have only a few matching characters. The ranking algorithm is Oraclecl
e8 can be found in several commercially available software packages.

Although there are many possible sequences of search algorithms, it is generally desirable to perform the narrowest or most specific search first. This sequence should continue the search algorithms according to the scope of each algorithm. For example, a search for a particular search string is the narrowest in range, but a search for the associated search string is more likely to return a wider range of matches, thus being broader in range. In one embodiment, a user can select a sequence of search algorithms to be applied to one or more data sets. In an alternative embodiment, the sequence of the search algorithm is preset by a system administrator.

An additional type of search algorithm not specifically shown in FIG. 2 is called "soundex.
dex) "or" sounds-l "
ike) "search, where the search string is checked against database records for similarity in speech. Various implementations of the soundex search algorithm are known in the art.

If any match is found as a result of the search algorithm, the subsequent search algorithm is skipped and the software 10 proceeds to display the result. If the software 10 was configured to add price and vendor information, this can also be added to the display.

In the preferred embodiment, as in FIG. 3, the display has several areas of information. The search string is shown in text box 300 and is preferably located in the upper left corner of the display. Matching items are displayed list 302
Shown in Display list 302 includes categories, manufacturer names, manufacturer part numbers, and descriptive attributes.

As an alternative to paging in a long product list, the user can narrow this list by selecting one of the categories. For example, FIG.
Indicates that two different categories of items were found: a pen and a pen refill. Since there are 29 matching items (only the first 10 are shown), button 3
By selecting 06, the list is narrowed to include only the pen and not the pen lead. As shown in FIG. 4, a new display is generated, and the updated display list 402 has 27 items that do not include the pen refill. In this embodiment, selecting a category becomes restrictive with respect to items in display list 302. This procedure is shown in a logic block 124 in FIG.

In general, items in the catalog have parameters that can be selected to restrict the display list 302. For example, the parameter for the pen can be "ink color", which has a value of blue, black, or red. If pen is selected as a category,
The display list 302 can be further limited to single color pens. This process is referred to herein as parameter refining.

In the preferred embodiment of the present invention, algorithm 200 has a list of synonyms for each attribute. The search algorithm replaces each search term with data and
It can be replaced with an appropriate synonym to match the record. Attributes are typically used as part of an algorithm for finding matches, and the use of synonyms for attributes provides additional flexibility in the range of search strings that produce meaningful matches.

Another way to increase the flexibility of the search algorithm is to allow natural adjectives in the search string to help select certain attributes. For example, if there is a category for computers and the category has an attribute for processor speed, the adjective "fastest" in the search string may be used to select the fastest computer.
Slower computers are removed from display list 302.

Yet another way to increase the flexibility of the search algorithm is to assign alternative roles to categories, such as, but not limited to, problem space and applications. . In this embodiment, the item found by the query is actually a predefined query, which generates a list of actual data records from the database. Therefore,
Selecting a category is equivalent to submitting a predefined query to the database.

The electronic billing system can provide links to further information about items in the database. Further information may be stored in database 14 or on a remote computer accessible via the Internet. For example, a button on a display screen can invoke a link to a web site containing relevant information.

Preferably, the items in the catalog are cross-referenced so that relevant items can be located quickly. For example, if the user selects a printer from a catalog, pressing a button on the display list can immediately locate the appropriate toner cartridge.

After the user has identified the desired product from database 14, software 10 may generate a purchase invoice with preformatted price, vendor and user information. Individual contract terms and price information can be accessed by software 10 and reflected in this bill. The invoice can be sent electronically to the vendor for processing.

It will be apparent to those skilled in the art that various modifications can be made within the scope of the invention described above. Accordingly, the present discovery is not to be considered as being limited to the particular examples selected for disclosure, but should include all changes and modifications that do not depart from the permitted scope of the invention. is there. Having thus described the present invention, what is desired to be assured and encompassed by the following claims is set forth in the appended claims.

[Brief description of the drawings]

FIG. 1 is a block diagram of one embodiment of an electronic catalog billing system.

FIG. 2 is a logical block diagram of a sequential search method according to the present invention.

FIG. 3 is a diagram of an example user interface, according to one embodiment of the invention.

[Explanation of symbols]

 10 Software 12 Application Server 14 Database 16 Storage Device 18 Workstation

 ──────────────────────────────────────────────────の Continued on the front page (72) Inventor James Michael Wilmsen United States, 80020, Colorado, Westminster Kalamas Court, 14731 (72) Inventor Christopher Wade Beer United States, 80026, La, Colorado Fayette Ginny Way 2440 F-term (reference) 5B075 ND20 PP22 PP30 PR10 QT05 QT10 UU38 5B082 GA08 GC04

Claims (3)

[Claims] 1. A method for selecting a data record in a computer readable database, the method comprising: specifying at least a first and a second search strategy and manipulating the data record to search for at least one match. Forming a search hierarchy consisting of the first search strategy followed by the second search strategy; inputting a search string including at least one search word into a user interface; Executing the search hierarchy, wherein the search string is applied by the first and second search strategies, and further comprising terminating the search hierarchy when at least one match is found; Displaying at least one match. 2. A system for selecting data records in a computer readable database, comprising: a plurality of data records in the database; and at least one of manipulating the data records to search for at least one match. An algorithm including a search hierarchy consisting of first and second search strategies, the first search strategy followed by the second search strategy, and a user interface having an input for a search string. Executing the algorithm causes the search string to be applied according to the algorithm, terminating the search hierarchy when the algorithm finds at least one match, and displaying the at least one match to the user interface. System to display on. 3. A method for selecting items in a computer readable electronic catalog, wherein each item has a classification and includes the step of specifying a plurality of subsets of data records in a database, wherein each subset is assigned to a different supplier. Corresponding, and further comprising specifying at least a first and a second search strategy, wherein each search strategy is preselected for manipulating at least one of the plurality of subsets to search for at least one match. Forming a search hierarchy consisting of the first search strategy followed by the second search strategy; and providing a search string comprising at least one search word to the user. Inputting to the interface; and executing the search hierarchy. Applying a string according to a sequence of the search strategies according to a preselected order of the suppliers, further comprising terminating the search hierarchy when the at least one match is found; Displaying.