Patent Grant
8239394
The present invention relates generally to search queries in search engines, such as Internet and Intranet search engines, and more specifically to restricting a database access operation in a search engine using Bloom filters.
Search engines provide a powerful tool for locating content in documents in a large database of documents, such as the documents on the Internet or World Wide Web (WWW), or the documents stored on the computers of an Intranet. The documents are located using an index of documents in response to a search query, consisting of one or more words, terms, keywords and/or phrases, that is submitted by a user.
Returning search results that closely match, or ideally, that do match a user's search query is a desired characteristic of search engines. This is especially the case in the current environment where there are ever increasing amounts of information. In addition, returning results promptly is also a desired characteristic.
The heterogeneous nature of search queries poses a challenge for processing them efficiently. Search queries are often free-form. The set of queries and the corresponding categories of information and/or content are not pre-defined. As a consequence, it may be difficult to efficiently access databases containing the index of documents and/or content corresponding to the index of documents. Furthermore, resources, such as memory, available for processing a respective set of operations, such as accessing a database, may be constrained by the sheer number of services that are running on the search engine at a given time. Pre-screening or simulating database access operations to determine if it is likely that corresponding information may be present in the database could improve the utilization of resources and the overall performance of search engines. There is a need, therefore, for search engines having such simulation capability.
A method of restricting a database access operation is disclosed. A plurality of Bloom filters corresponding to a plurality of fields in a database are applied to two or more terms in a record to produce a first plurality of outputs. The record may correspond to a query. Extraneous terms may be removed from the record. The access operation to the database is enabled if the first plurality of outputs correspond to a positive result.
Two or more terms in the record may be sorted to produce a sorted record. A first Bloom filter may be applied to the sorted record to produce a sorted output. The access operation to the database may be enabled if the sorted output corresponds to a positive result.
The database may include factual information. The plurality of fields in the database may include objects, attributes and values.
In some embodiments, first and second subsets of the terms in the query are identified. A first Bloom filter in the plurality of Bloom filters is applied to the first subset to produce a first output. A second Bloom filter in the plurality of Bloom filters is applied to the second subset to produce a second output. The access operation is enabled when the first output and the second output both correspond positive results.
In some embodiments, a plurality of subsets of the terms in the query are identified. One or more of the Bloom filters are applied to each of the subsets to produce a second plurality of outputs. The access operation is enabled when the second plurality of outputs correspond to at least one of a plurality of success patterns.
In some embodiments, a normalization operation is applied to the query to produce a normalized set of terms. An additional Bloom filter is applied to each of the terms in the normalized set of terms to produce an additional output. The access operation is enabled in further accordance with the additional output corresponding to a positive result.
In some embodiments, the terms in the record are parsed in a plurality of ways to produce a plurality of parsed sets of terms. At least two of the Bloom filters in the plurality of Bloom filters are applied to the plurality of parsed sets of terms to produce a third plurality of outputs. The access operation is enabled in accordance with the third plurality of outputs corresponding to a positive result. The two Bloom filters may respectively correspond to an attribute field and a value field of a database.
In some embodiments, a first Bloom filter in the plurality of Bloom filters is applied to one or more terms in the record to produce a third output. Remaining terms in the record are sorted. A second Bloom filter in the plurality of Bloom filters is applied to the sorted remaining terms in the record to produce a fourth output. The access operation is enabled in further accordance with the third output and the fourth output corresponding to positive result. A first field corresponding to the first Bloom filter may include attributes. A second field corresponding to the second Bloom filter may include values. The one or more terms in the record may include a first term, a first two terms, a last term and/or a last two terms.
For a better understanding of the invention, reference should be made to the following detailed description taken in conjunction with the accompanying drawings, in which:
Like reference numerals refer to corresponding parts throughout the drawings.
Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.
Restricting a database access operation using a plurality of Bloom filters is described. The Bloom filters contain one or more hashing functions and a corresponding filter array. A respective filter array contains entries corresponding to outputs from the one or more hashing functions when applied to respective terms input to the respective Bloom filter. The Bloom filters characteristics are selected based on a size of the database, a size of the filter array, a number and type of hashing functions and a pre-determined acceptable false positive rate. The Bloom filters are used to simulate an access operation based on a record. As explained in more detail below, if an output from one or more of the Bloom filters corresponds to a positive result the access operation is enabled.
In some embodiments, the record is a search query containing two or more words, terms, keywords and/or phrases. Henceforth, contents of a search query are referred to as terms. The search query may correspond to content located at one or more document locations. The one or more documents locations may correspond to one or more web sites and/or one or more web pages. The one or more document locations may include one or more uniform resource locators (URLs). The one or more document locations may be on an Intranet and/or the Internet, which is also referred to as the World Wide Web (WWW).
In an exemplary embodiment, the search query is a request for factual information, e.g., What is the capital of Spain? Such a request for factual information is henceforth called a fact query. A respective fact query may be processed by a regular or generic search engine and/or a facts search engine. Processing by the generic search engine and the facts search engine may occur sequentially or substantially simultaneously, i.e., substantially in parallel.
The facts search engine may include a facts database or facts repository. The facts repository may include a plurality of fields, including a name field, an attribute field and a value field. Attributes and values in the facts repository may be arranged in pairs, i.e., as attribute-value pairs.
The plurality of Bloom filters may be used to simulate an access operation to the facts repository based on a facts query. This will allow facts queries that do not correspond to content in the facts repository to be dropped, i.e., to block or not enable an access operation to the facts repository for such facts queries. While there is a small probability that outputs from the Bloom filters provide an erroneously positive result for a respective facts query, overall the Bloom filters offer a fast way to pre-screen or simulate access operations to improve resource allocation (for resources such as CPU cycles or search engine bandwidth) and performance of the facts search engine by reducing a number of facts queries that are processed.
While
There are a variety of approaches for using the plurality of Bloom filters, such as the Bloom Filters 314 (
The database in embodiments 400, 450 and 460 may include factual information. The database may include the facts repository. Each of the Bloom filters may correspond to a name or objects field, an attribute field, a value field in the facts repository and/or a term field. The term field may include terms that are included in one or more of the other fields in the database.
In method 460, the term or terms in the record to which the first Bloom filter is applied may include a first or leading term in the record, the first two terms in the record, the last two terms in the record and/or the last term in the record. In operation 430, the first Bloom filter may correspond to an attributes field in the database. In operation 434, the second Bloom filter may correspond to a values field.
The Bloom filters may have a false positive rate of around 2% or less, and the Bloom filters may be allocated up to 20 MB of memory with which to perform simulations of access operations to the facts repository in the facts search engine. In an exemplary embodiment, the Bloom filters may be constructed using 2 bytes (i.e., 16 bits) per entry in a respective Bloom filter. Thus, in a term Bloom filter (discussed below, for example, with reference to
When a search query is received, it is processed by the facts search engine and the generic search engine, or in related support hardware in the system (as discussed below with reference to
Each remaining term in the search query is processed using a term Bloom filter corresponding to the term field in the facts repository. If any of the terms in the search query does not appear in a term Bloom filter array, i.e., if the output from the term Bloom filter does not correspond to the positive result, the access operation to the facts repository may not be enabled. The terms in the search query are sorted, for example, alphabetically, and a resulting string is processed using a name Bloom filter that corresponds to the name field in the facts repository. If the output from the name Bloom filter corresponds to the positive result, the access operation to the facts repository may be enabled.
A first or leading term in the search query is processed using an attribute Bloom filter. Remaining terms in the search query are sorted, for example, alphabetically, and a resulting string is processed using a value Bloom filter. If outputs from the attribute Bloom filter and the value Bloom filter correspond to the positive result, the access operation to the facts repository may be enabled.
The operations discussed in the preceding paragraph may be repeated a plurality of times, where the attribute Bloom filter during an instance of the operations is applied to the first two terms in the search query, the last two terms in the search query and/or the last term in the search query. In addition, the value filter may be applied to the remaining terms in the search query.
If the simulation does not enable the access operation to the facts repository, the generic search engine provides results in response to the search query. If the access operation is enabled, results from the facts search engine and/or the generic search engine may be provided. Results may include a ranked set of documents, snippets of content in a ranked set of documents and/or a ranked set of document locations or links. Results from the facts search engine and the generic search engine may be displayed in separate lists and/or in a common list.
As an illustration, consider a search query of “Bob Smith date of birth.” Normalizing the search query may remove the term “of.” The normalized search query becomes “Bob Smith date birth.”
“Bob,” “Smith,” “date” and “birth” may each be tested against the term Bloom filter. If the output from the term Bloom filter is not a positive result, i.e., if any term produces a negative result, the access operation to the facts repository is not enabled. If all the terms in the normalized search query produce a positive result when tested against the term Bloom filter, simulation of the search query continues.
Next, the terms of the normalized search query are sorted, for example alphabetically to produce “birth Bob date Smith,” and the resulting string is tested against the name Bloom filter. If the result is positive the access operation is enabled, and if the result is negative and the normalized search query has more than one term (which it does, in this example), simulation of the access operation continues.
The terms in the normalized search query may be parsed in a variety of ways and processed using two of more of the Bloom filters. In this example, the terms in the normalized query are parsed in four ways, with each parsing producing two sets of terms: Set1 and Set2. The four parsings are: 1) Set1: first term, Set2: remaining terms in sorted order; 2) Sett: first two attribute terms, Set2: remaining terms in sorted order; 3) Set1: last term, Set2: remaining terms in sorted order; and 4) Set1: last two attribute terms, Set2: remaining terms in sorted order. For the normalized search query “Bob Smith date birth,” the four parsings are: 1) Set1:“Bob,” Set2:“birth date Smith”; 2) Set1: “Bob Smith”, Set2: “birth date”; 3) Set1: “birth”, Set2: “Bob date Smith”; 4) Set1: “birth date”, Set2: “Bob Smith”.
For each of these four pairs of sets, the Attribute Bloom filter is applied to Sell to produce a first result and the value Bloom filter is applied to Set2 to produce a second result. As a result, there can be four sets of results. If both results (i.e., the first and second results) in any of these four sets are positive, then the search query is passed to the fast search engine for processing. Whenever a pair of positive results is generated, the simulation is stopped and the search query is passed to the facts search engine for processing. If none of the four pairs of results are both positive, the search query is not sent to the facts search engine for processing.
In this example, the application of the value Boom filter to “Bob Smith” and the attribute filter to “birth date” will both produce positive results, and therefore the search query is passed (using parsing option #4 described above) to the facts search engine. In this example, none of the other pairs of results are both positive, but only one pair of positive results is required for the search query to be passed to the facts search engine for processing. In other embodiments, additional parsing patterns and/or additional Bloom filter tests could be used when determining whether to send the search query to a facts search engine for processing.
In some embodiments, one or more of these ways of parsing the search query may be used. In some embodiments, the access operation to the facts repository is enabled if two of the Bloom filters in a set of Bloom filters including the name Bloom filter, the attribute Bloom filter and the value Bloom filter have an output corresponding to the positive result. In embodiments where the search query is a name query, i.e., the output from the name Bloom filter corresponds to the positive result, the access operation to the facts repository may be enabled.
Attention is now given to hardware and systems that may utilize and/or implement the simulation of a database access operation and/or the embodiments 400, 450, and 460 of the methods discussed above.
Note that the search engine system 500 can be used as an Internet search engine, for locating documents on the WWW and/or as an Intranet search engine, for locating documents stored on servers or other hosts within an Intranet. In addition, the methodology described herein is applicable to implementations where only portions of documents, such as titles and abstracts, are stored in the database of the search engine system 500.
The search engine system 500 may include multiple data centers each housing a backend. The data centers are generally widely dispersed from one another, such as across the continental United States. Search queries submitted by users at one of the clients 510 to the search engine system 500 are routed to an appropriate backend as part of the Domain Name System (DNS), based on current load, geographic locality and/or whether that data center is operating.
Each backend preferably includes multiple query servers, such as query server 516, coupled to a communications network 514 via a client communication module 518. The communications network 514 may be the Internet, but may also be any local area network (LAN) and/or wide area network (WAN). In some embodiments, each query server 516 is a Web server that receives search query requests and delivers search results in the form of web pages via HTTP, XML or similar protocols. Alternatively, if the query server 516 is used within a LAN, i.e., internally and not by the public, it may be an Intranet server. In essence, the query servers, such as query server 516, are configured to control the search process, including searching a document index, analyzing and formatting the search results.
Each backend may also include a query receipt, processing and response module 520 for managing the processing of search queries by multiple query processing controllers, such as query processing controller 522 in the generic search and facts query processing controller 538 in the facts search engine, that are coupled to the query server 516. Each query processing controller 522 may be coupled to a cache 530, an inverse document index 526 and a document database 528. Each query processing controller 522 is configured to receive requests from one of the query servers, such as the query server 516, and transmit the requests to the cache 530, the inverse document index 526 and the document database 528.
The cache 530 is used to increase search efficiency by temporarily storing previously located search results. The efficiency and cost of performing a search is dependent on a number of factors, such as the various combinations of terms and/or keywords used in the search query, the length of time spent on the search and the number of documents indexed. The average response time and cost of a search is lowered by storing the search results of previously requested search queries in the temporary storage or cache 530 of the search engine system 500. One of the benefits of keeping a cache of commonly searched information is that it ensures a fast response to subsequent search queries for the same information, as it takes less time to retrieve the information from local memory. Another benefit of caching commonly searched information is to reduce the cost of servicing multiple search queries requesting the same information. When the result of a search is cached, there is no need to perform another search of the document database 528, which may be distributed over multiple disks and/or memory arrays on multiple remote computer servers. Retrieving such information from the multiple disks and/or memory arrays in multiple remote computer servers requires significantly more time and resources than retrieving the information from the cache 530. Instead, an index contains a pointer to a cache entry corresponding to a current search query.
The search rank values for the documents in the search results are conveyed to the query processing controller 522 and/or the query server 516, and are used to construct an ordered search result list. Once the query processing controller 522 constructs the ordered search result list, the query processing controller 522 transmits to the document database 528 a request for snippets of an appropriate subset of the documents in the ordered search list. For example, the query processing controller 522 may request snippets for the first fifteen or so of the documents in the ordered search result list. The document database 528 constructs snippets based on the search query, and returns the snippets to the query processing controller 522. The query processing controller 522 then returns a list of located documents and snippets back to the query server 516. In some embodiments, the snippets are stored in the cache server 530 along with the cached results. As a result, in these embodiments the query processing controller 522 only requests snippets for documents, if any, for which it is unable to obtain valid cached snippets from the cache server 530.
The facts query processing controller 538 may similarly manage the processing of a current search query. Prior to accessing one or more instances of a partitioned facts index 542, one or more instances of a partitioned facts repository 544 and/or a facts cache 546, the facts query processing controller 538 may utilize Bloom filters 540 to simulate an access operation using an embodiment of the methods described previously. In this way, the facts query processing controller 538 may improve resource allocation and the efficiency of the facts search engine. In another embodiment, the Bloom filters 540 are located in the query server 516, which utilizes the Bloom filters 540 to simulate an access operation using an embodiment of the methods described previously.
The query server 516 may process results provided by the query processing controller 522 (i.e., results from the generic search engine) and/or the facts query processing controller 538 (i.e., results from the facts search engine). The query server 516 may determine if results from the facts search engine correspond to a respective search query (e.g., by testing the results against reliability criteria and/or relevance criteria) and may determine which results are provided to a client, such as client 510_1, and a format of the results.
The memory 622 may include high-speed random access memory, including solid state or integrated circuit memory devices such as dynamic random access memory and/or flash memory devices, and/or non-volatile memory, such as one or more magnetic disk storage devices, optical storage devices, and/or static memory. The memory 622 may store an operating system 624, such as LINUX, UNIX or WINDOWS, that includes procedures (or a set of instructions) for handling basic system services and for performing hardware dependent tasks. The memory 622 may also store communication procedures (or a set of instructions) in a network communication module 626. The communication procedures are used for communicating with clients, such as the clients 510 (
The memory 622 may also store the query server 516 (or a set of instructions), query processing controller 522 (or a set of instructions), facts processing controller 538 (or a set of instructions), facts index 542, facts repository 544 and answer identification (or a set of instructions) 636. The query server 516 may include the following elements, or a subset or superset of such elements: the client communication module 518 and the query receipt, processing and response module 520. The facts processing controller 538 may include Bloom filters 540, a term identification module 630, a normalization module 632 and a sorting module 634.
Although
The memory 722 may include high-speed random access memory, such as DRAM, SRAM, DDR RAM or other random access solid state memory devices, and may include non-volatile memory, such as one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or other non-volatile solid state storage devices. Memory 722 may optionally include one or more storage devices remotely located from the CPU(s) 710. The memory 722 may store an operating system 724, such as LINUX, UNIX or WINDOWS, that includes procedures (or a set of instructions) for handling basic system services and for performing hardware dependent tasks. The memory 722 may also store communication procedures (or a set of instructions) in a network communication module 726. The communication procedures are used for communicating with a search engine.
The memory may also include a browser or browser tool module 728 (or a set of instructions) and a search assistant module 730 (or a set of instructions). The search assistant module 730 may include an entry and selection monitoring module 732 for monitoring user input, a transmission module 734 for sending a search query, a search results receipt module 736 for receiving search results, and a display module 738 for displaying search results as well as web pages, html or XML documents, and/or other documents.
In embodiments where the client system 700 is coupled to a local server computer, one or more of the modules and/or applications in the memory 722 may be stored in a server computer at a different location than the user.
Each of the above identified modules and applications corresponds to a set of instructions for performing one or more functions described above. These modules (i.e., sets of instructions) need not be implemented as separate software programs, procedures or modules. The various modules and sub-modules may be rearranged and/or combined. The memory 722 may include additional modules and/or sub-modules, or fewer modules and/or sub-modules. For example, the search assistant module 730 may be integrated into the browser/tool module 728. The memory 722, therefore, may include a subset or a superset of the above identified modules and/or sub-modules.
Some of the data structures used in search engines, such as the search engine 600 (
As discussed previously,
The foregoing descriptions of specific embodiments of the present invention are presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed. Rather, it should be appreciated that many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated.
| Number | Name | Date | Kind |
|---|---|---|---|
| 4888690 | Huber | Dec 1989 | A |
| 4899292 | Montagna et al. | Feb 1990 | A |
| 5475819 | Miller et al. | Dec 1995 | A |
| 5544051 | Senn et al. | Aug 1996 | A |
| 5560005 | Hoover et al. | Sep 1996 | A |
| 5574898 | Leblang et al. | Nov 1996 | A |
| 5778378 | Rubin | Jul 1998 | A |
| 5815415 | Bentley et al. | Sep 1998 | A |
| 5832479 | Berkowitz et al. | Nov 1998 | A |
| 5870739 | Davis, III et al. | Feb 1999 | A |
| 5905980 | Masuichi et al. | May 1999 | A |
| 5946692 | Faloutsos et al. | Aug 1999 | A |
| 5963940 | Liddy et al. | Oct 1999 | A |
| 6006221 | Liddy et al. | Dec 1999 | A |
| 6014661 | Ahlberg et al. | Jan 2000 | A |
| 6026388 | Liddy et al. | Feb 2000 | A |
| 6029195 | Herz | Feb 2000 | A |
| 6101515 | Wical et al. | Aug 2000 | A |
| 6105020 | Lindsay et al. | Aug 2000 | A |
| 6105030 | Syed et al. | Aug 2000 | A |
| 6216138 | Wells et al. | Apr 2001 | B1 |
| 6222540 | Sacerdoti | Apr 2001 | B1 |
| 6263328 | Coden et al. | Jul 2001 | B1 |
| 6263335 | Paik et al. | Jul 2001 | B1 |
| 6304864 | Liddy et al. | Oct 2001 | B1 |
| 6311189 | deVries et al. | Oct 2001 | B1 |
| 6326962 | Szabo | Dec 2001 | B1 |
| 6327574 | Kramer et al. | Dec 2001 | B1 |
| 6377943 | Jakobsson | Apr 2002 | B1 |
| 6480194 | Sang'udi et al. | Nov 2002 | B1 |
| 6519631 | Rosenschein et al. | Feb 2003 | B1 |
| 6529900 | Patterson et al. | Mar 2003 | B1 |
| 6584464 | Warthen | Jun 2003 | B1 |
| 6606659 | Hegli et al. | Aug 2003 | B1 |
| 6609123 | Cazemier et al. | Aug 2003 | B1 |
| 6629097 | Keith | Sep 2003 | B1 |
| 6643641 | Snyder | Nov 2003 | B1 |
| 6718324 | Edlund et al. | Apr 2004 | B2 |
| 6832218 | Emens et al. | Dec 2004 | B1 |
| 6850896 | Kelman et al. | Feb 2005 | B1 |
| 6873982 | Bates et al. | Mar 2005 | B1 |
| 6885990 | Ohmori et al. | Apr 2005 | B1 |
| 6928436 | Baudel | Aug 2005 | B2 |
| 6961723 | Faybishenko et al. | Nov 2005 | B2 |
| 6968343 | Charisius et al. | Nov 2005 | B2 |
| 7013308 | Tunstall-Pedoe | Mar 2006 | B1 |
| 7031955 | de Souza et al. | Apr 2006 | B1 |
| 7043521 | Eitel | May 2006 | B2 |
| 7100083 | Little et al. | Aug 2006 | B2 |
| 7146538 | Johnson et al. | Dec 2006 | B2 |
| 7158983 | Willse et al. | Jan 2007 | B2 |
| 7669115 | Cho et al. | Feb 2010 | B2 |
| 20020055954 | Breuer | May 2002 | A1 |
| 20020065814 | Okamoto et al. | May 2002 | A1 |
| 20020065815 | Layden | May 2002 | A1 |
| 20020128818 | Ho et al. | Sep 2002 | A1 |
| 20020154175 | Abello et al. | Oct 2002 | A1 |
| 20020173984 | Robertson et al. | Nov 2002 | A1 |
| 20030005036 | Mitzenmacher | Jan 2003 | A1 |
| 20030069880 | Harrison et al. | Apr 2003 | A1 |
| 20030097357 | Ferrari et al. | May 2003 | A1 |
| 20030115485 | Milliken | Jun 2003 | A1 |
| 20030120373 | Eames | Jun 2003 | A1 |
| 20030120644 | Shirota | Jun 2003 | A1 |
| 20030120654 | Edlund et al. | Jun 2003 | A1 |
| 20030120659 | Sridhar | Jun 2003 | A1 |
| 20030154071 | Shreve | Aug 2003 | A1 |
| 20030158855 | Farnham et al. | Aug 2003 | A1 |
| 20030195872 | Senn | Oct 2003 | A1 |
| 20030208665 | Peir et al. | Nov 2003 | A1 |
| 20030217052 | Rubenczyk et al. | Nov 2003 | A1 |
| 20040015566 | Anderson et al. | Jan 2004 | A1 |
| 20040030731 | Iftode et al. | Feb 2004 | A1 |
| 20040107125 | Guheen et al. | Jun 2004 | A1 |
| 20040122844 | Malloy et al. | Jun 2004 | A1 |
| 20040125137 | Stata et al. | Jul 2004 | A1 |
| 20040167909 | Wakefield et al. | Aug 2004 | A1 |
| 20040220904 | Finlay et al. | Nov 2004 | A1 |
| 20040236655 | Scumniotales et al. | Nov 2004 | A1 |
| 20040255237 | Tong | Dec 2004 | A1 |
| 20040267700 | Dumais et al. | Dec 2004 | A1 |
| 20050022009 | Aguilera et al. | Jan 2005 | A1 |
| 20050033803 | Vleet et al. | Feb 2005 | A1 |
| 20050039033 | Meyers et al. | Feb 2005 | A1 |
| 20050050016 | Stanoi et al. | Mar 2005 | A1 |
| 20050055327 | Agrawal et al. | Mar 2005 | A1 |
| 20050057566 | Githens et al. | Mar 2005 | A1 |
| 20050060277 | Zlatanov et al. | Mar 2005 | A1 |
| 20050076012 | Manber et al. | Apr 2005 | A1 |
| 20050086520 | Dharmapurikar et al. | Apr 2005 | A1 |
| 20050120004 | Stata et al. | Jun 2005 | A1 |
| 20050187898 | Chazelle et al. | Aug 2005 | A1 |
| 20050216464 | Toyama et al. | Sep 2005 | A1 |
| 20050219929 | Navas | Oct 2005 | A1 |
| 20050256825 | Dettinger et al. | Nov 2005 | A1 |
| 20050268212 | Dagel | Dec 2005 | A1 |
| 20060004851 | Gold et al. | Jan 2006 | A1 |
| 20060020582 | Dettinger et al. | Jan 2006 | A1 |
| 20060047838 | Chauhan | Mar 2006 | A1 |
| 20060085386 | Thanu et al. | Apr 2006 | A1 |
| 20060085465 | Nori et al. | Apr 2006 | A1 |
| 20060149700 | Gladish et al. | Jul 2006 | A1 |
| 20060173824 | Bensky et al. | Aug 2006 | A1 |
| 20060206508 | Colace et al. | Sep 2006 | A1 |
| 20060224582 | Hogue | Oct 2006 | A1 |
| 20070022085 | Kulkarni | Jan 2007 | A1 |
| 20070055656 | Tunstall-Pedoe | Mar 2007 | A1 |
| 20070067108 | Buhler et al. | Mar 2007 | A1 |
| 20070179965 | Hogue et al. | Aug 2007 | A1 |
| 20070203867 | Hogue et al. | Aug 2007 | A1 |
| 20070203868 | Betz | Aug 2007 | A1 |
| 20070271249 | Cragun et al. | Nov 2007 | A1 |
| 20080005064 | Sarukkai | Jan 2008 | A1 |
| 20080097958 | Ntoulas et al. | Apr 2008 | A1 |
| 20080209444 | Garrett et al. | Aug 2008 | A1 |
| Number | Date | Country |
|---|---|---|
| 11-265400 | Sep 1999 | JP |
| 2002-157276 | May 2002 | JP |
| 2002-540506 | Nov 2002 | JP |
| 2003-281173 | Oct 2003 | JP |
| WO 2004114163 | Dec 2004 | WO |
