US 7,428,522 B1Grant
Real-time search term popularity determination, by search origin geographic location
Issue Date:2008-09-23
•5 Claims
•7 Drawing Sheets
Abstract
Information is generated indicative of frequency of search terms presented to at least one online search service. As event indications, indicative of user interaction generally with front end servers, are being provided for persistent storage, ones of the event indications that are indicative of search events are detected. The detected ones of the search event indications are processed and it is determined, based at least in part thereon, by location, frequency data indicative of a frequency of each of a plurality of search terms presented to the at least one online search service. An indication of at least some of the frequency data is caused to be associated with indications of locations to which the frequency data corresponds. For example, the frequency data may be displayed superimposed on a map.
Metadata
Assignee
- Yahoo! Inc.
Inventor
- Vijay Raghunathan
Application Information
Application Number:US 11/862,967
Filing Date:2007-09-27
Priority Date:2007-09-27
Art Unit:2167
Classifications
IPC:
G06F17/30G06F7/00
Patent Drawings (7 sheets)
Description
Background
[0001] It is known to process a search index, of an online search service, to determine a level of popularity of query terms made in a given time period, such as during a given day. Conventionally, such popularity determinations are made based on transaction data in a data warehouse, and it can take up to two to three days for the search query data in the data warehouse to be processed and the search index (with information about the popularity level of query terms) to be built and cleansed. Furthermore, the resolution of the popularity determinations (for example, temporally and geographically) may be limited.
Summary
[0002] In accordance with an aspect of the invention, information is generated indicative of frequency of search terms presented to at least one online search service. As event indications, indicative of user interaction generally with front end servers, are being provided for persistent storage, event indications that are indicative of search events are detected. The detected search event indications are processed and it is determined, based at least in part thereon, by event origin location, frequency data indicative of a frequency of each of a plurality of search terms presented to at least one online search service. An indication of at least some of the frequency data is caused to be associated with indications of event origin locations to which the frequency data corresponds. For example, the frequency data may be displayed superimposed on a map including event origin locations.
[0003] Processing the detected ones of the search event indications and determining based at least in part thereon, by event origin location, frequency data indicative of a frequency of each of a plurality of search terms presented to the at least one online search service may include updating the determination of frequency data based on newly processed detected search event indications. Updating the determinations of frequency data based on newly processed detected search event indications may, for example, include applying an algorithm that automatically decays frequency values over time.
Brief Description of the Drawings
[0004] FIG. 1 , which does not illustrate the present invention, illustrates an architecture of a system in which front end web servers, including front end web servers handling search events, are producing event data based on incoming user requests, including search queries each comprising at least one search term.
[0005] FIG. 2 schematically illustrates event data provided to the data collectors are further provided to a data warehouse via a “data highway,” and a filter selects search event data.
[0006] FIG. 3 illustrates an example architecture of the statistics generation block of FIG. 2 .
[0007] FIG. 4 illustrates schematically how various arrays are “dumped” to a data structure accessible for display.
[0008] FIG. 5 illustrates an example of a display that may be generated based on generated statistics.
[0009] FIG. 6 is a flowchart of a method in accordance with an embodiment of the invention, to cause determination and display of the level of popularity of query terms provided to one or more search services.
Detailed Description
[0010] The inventors have realized that it can be useful for the level of popularity of query terms to be determined relatively quickly. For example, the determination may be useful in the process of marketing the keywords for advertisement targeting and/or other monetization. For example, it may be possible to target a “fresh” keyword or keywords before the market for that keyword is saturated. The inventors have additionally realized that, if the popularity determination can be resolved geographically and/or temporally, this may also provide advantages in marketing a keyword or keywords.
[0011] In accordance with an aspect, a method/system is described to determine high-frequency keywords in a search-stream, with a low latency. Additionally, the stream may be partitioned geographically (e.g., by city or other event origin location) and/or temporally (e.g., by hour or by time of day such as morning, afternoon, evening, etc.), allowing a user of the determinations to discern geographic and temporal differences in monetizing specific keywords. Search-events from front-end web-servers handling the search queries are processed in real-time. The latency of the frequency determinations may be less than one hour, or even under five minutes. Indications of the high-frequency keywords may be displayed, for example, relative to a map to allow for easy navigation.
[0012] FIG. 1 , which does not illustrate the present invention, illustrates an architecture of a system in which front end web servers FEa 102a, FEb 102b, FEc 102c, . . . , FEx 102x, including front end web servers handling search events, are producing event data 105 based on incoming user requests 103, including search queries each comprising at least one search term. The event data 105 is provided to data collectors DC1 108(1) and DC2 108(2) via paths Pa 106a, Pb 106b, Pc 106c and Pd 106d. In general, there may be numerous front end web servers, data collectors and paths; a small number are shown in FIG. 1 and throughout this patent description for simplicity of illustration. The particular paths may be determined according to a path configuration 104, for example, as described in U.S. patent application Ser. No. 11/734,067, filed on Apr. 11, 2007. U.S. patent application Ser. No. 11/734,067 is incorporated by reference at least for its disclosure of methods to determine path configurations.
[0013] The data collectors may be, for example, computers or computer systems in one or more data centers. A data center is a collection of machines (data collector machines) that are co-located (i.e., physically proximally-located). The data centers may be geographically dispersed to, for example, minimize latency of data communication between front end web servers and the data collectors. Within a data center, the network connection between machines is typically fast and reliable, as these connections are maintained within the facility itself. Communication between front end web servers and data centers, and among data centers, is typically over public or quasi-public networks (i.e., the internet).
[0014] Referring now to FIG. 2 , the event data provided to the data collectors DC1 108(1) and DC2 108(2) via paths Pa 106a, Pb 106b, Pc 106c and Pd 106d are further provided to a data warehouse 202 via what may be thought of as a “data highway” 204. A filter 206 selects events that are from front end web servers that are search web servers.
[0015] For example, every event may be indicated by an event record that includes fields whose contents characterize the event. For example, an event record may include a field whose contents identify a “host name” or “space id” corresponding to a front end server that generated the event. A “space id” is a unique key that identifies the page contents and category. The contents of these event record fields may thus be highly indicative of whether the event is a search event. Other fields in an event record may include a time stamp, the URL accessed by a user, and an IP address via which a computing device accesses a search service (e.g., a computing device that executes a browser to access the search service). Other fields may be dependent on the type of event such as, for example, a field for a query term if the event is a search event.
[0016] We now discuss the data highway 204 in greater detail. The data highway 204 is a data pipeline via which a stream of events generated by data producers are provided to data consumers. The pipeline has been denoted as a “highway,” because there are multiple off ramps but, in one example, the pipeline is unidirectional. The highway may comprise multiple stages. Each stage forwards the events toward the data consumers, which in turn can be another stage, so the stages may be chained together. Additionally, each stage may include at least one filter to filter an event based on characteristics of the event. For instance, an event filter may be configured to “only give events that have been generated by data producer Y” or “only give events that occurred between 12:30 and 1:00 PM.” Event filters may be combined together using Boolean logic (e.g., give events that come from data producer Y that occurred between 12:30 and 1:00 PM). In general, events that are filtered may also continue along their original path to the data warehouse 202.
[0017] For events determined by the filter 206 (e.g., from the host name or via some other means) to be a search event or highly likely to be a search event, statistics are generated (208) in real time on the basis of some characteristic evidenced by the events. More particularly, statistics are generated to correlate events having particular characteristics with the search query terms presented to the search service. An indication of the generated statistics is caused to be displayed 210.
[0018] In the FIG. 2 example, the statistics are generated based on geographic locations associated with the origins of the events. That is, statistics are generated to correlate events of originating from particular geographic locations or regions with the search query terms of those events. The geographic locations may be determined, for example, by correlating the IP address via which a computing device accesses the search service, as indicated in an event record, to a known or discerned location associated with that IP address. As another example, a geographic location associated with an event origination may be determined based on one or more cookies in the event record for that event. For example, for an event corresponding to a logged-in Yahoo! user, the geographic location for that event may be determined based on known demographic information maintained by Yahoo! and corresponding to that logged-in Yahoo! user.
[0019] FIG. 3 illustrates an example architecture of the statistics generation block 208 of FIG. 2 . In block 302, the search events are partitioned by geographic location or other characteristic indicated in the search event and which is to be correlated to the search query terms to generate the statistics. Thus, for example, if the statistics are to be generated to correlate particular geographic locations to the search query terms presented to the search service from those geographic locations, then the events are partitioned by geographic location. (For example, a geolocation targeting call, to a specialized geolocation program, may be employed to determine the latitude & longitude associated with the search event based on the IP address associated with the search event.) Thus, for example, the processing of events may be partitioned into distinct disjoint sets to distribute the statistics generation processing. Partitioning based on geography can be useful given that, in general, events from multiple geographic locations are not processed to determine the most popular search terms in any one particular geographic location. The partitioning facilitates scaling of the statistic generation processing for greater numbers of events and/or for additional geographic locations or other event characteristics.
[0020] Partitions 304a, 304b, 304c, . . . , 304x generate statistics regarding frequency of search terms correlated to search event characteristics such as geographic location. It is noted that while, in some examples, each partition may handle statistics generation for one or more search event characteristic values (and/or aggregation of search event characteristic values), in general, statistics generation for a particular one or more search event characteristics values (or aggregation of search event characteristic values) is not distributed across multiple partitions. Thus, for example, partition 304a may generate statistics 306a1, 306a2 and 306a3 for three respective search event characteristic values (and/or aggregation of search event characteristic values). Similarly, partition 304b may generate statistics 306b1 and 306b2 for two respective search event characteristic values (and/or aggregation of search event characteristic values); partition 304c may generate statistics 306c1 and 306c2 for two respective search event characteristic values (and/or aggregation of search event characteristic values); and partition 304x may generate statistics 306x for one search event characteristic value (and/or aggregation of search event characteristic values).
[0021] The statistics 306a1, 306a2 and 306a3 (generated by partition 304a); 306b1 and 306b2 (generated by partition 304b); 306c1 and 306c2 (generated by partition 304c) and 306x1 (generated by partition 304x) are generated and maintained locally by each respective partition 304a, 304b, 304c, . . . , 304x. The generated statistics are occasionally (e.g., periodically) “dumped” to a location 308 accessible for display 210.
[0022] In accordance with one example, the statistics generation employs a traffic analysis algorithm that generates, on a rolling basis, an indication of the most popular search query terms based on the events presented to it. The traffic analysis algorithm may be configured to account not only for frequency, but also for temporality, in the statistics generation.
[0023] As a specific example, the algorithm may be such that frequency values for search query terms automatically decay over time. A simplistic example of one particular implementation is now described. In the example, an array is used by each partition 304 to accumulate statistics regarding frequency of search terms represented by events presented to that partition. In particular, each item of the array indicates one search query term and a frequency associated with that search term. In the example, it is assumed that the array is of size three and, initially, the items of the array are empty. If the following search term data is sent into the partition:
[0024] {“a”, “b”, “c”, “a”, “a”, “d”, “d”, “b”}
[0025] the first three letters will be added (step 1) since, as each of the first three letters is processed, the array is not full yet. The column “current state of array” includes each keyword and the current calculated frequency for that keyword. The star is the moving pointer to a ‘current’ index in the array). After step 1, the “current” index is to the first item in the array, containing “a.”
[0026]
At step 2, the second “a” has been processed. Since the “current” index is already pointing to an item in the array that contains “a,” the frequency for “a” is incremented to two at step 2. At step 3, the third “a” has been processed. Like at step 2, since the “current” index is already pointing to an item in the array that contains “a,” the frequency for “a” is incremented to three at step 3.
| Step | Data already processed | current state of array | data to be processed |
|---|---|---|---|
| 1 | {‘a’, ‘b’, ‘c’} | {*‘a’->1, ‘b’->1, ‘c’->1} | {‘a’, ‘a’, ‘d’, ‘d’, ‘b’} |
| 2 | {‘a’, ‘b’, ‘c’, ‘a’} | {*‘a’->2, ‘b’->1, ‘c’->1} | {‘a’, ‘d’, ‘d’, ‘b’} |
| 3 | {‘a’, ‘b’, ‘c’, ‘a’, ‘a’} | {*‘a’->3, ‘b’->1, ‘c’->1} | {‘d’, ‘d’, ‘b’} |
| 4 | {‘a’, ‘b’, ‘c’, ‘a’, ‘a’, ‘d’} | {‘a’->2, *‘b’->1, ‘c’->1} | {‘d’, ‘b’} |
| 5 | {‘a’, ‘b’, ‘c’, ‘a’, ‘a’, ‘d’, ‘d’} | {‘a’->2, ‘d’->1, *‘c’->1} | {‘b’} |
| 6 | {‘a’, ‘b’, ‘c’, ‘a’, ‘a’, ‘d’, ‘d’, ‘b’} | {*‘a’->2, ‘d’->1, ‘b’->1} | { } |
At step 2, the second “a” has been processed. Since the “current” index is already pointing to an item in the array that contains “a,” the frequency for “a” is incremented to two at step 2. At step 3, the third “a” has been processed. Like at step 2, since the “current” index is already pointing to an item in the array that contains “a,” the frequency for “a” is incremented to three at step 3.
[0027] At step 4, a “d” has been processed. Since the “current” index points to an item in the array that contains “a” and does not contain “d,” the frequency for “a” is decremented and the “current” index is moved to point to the next item in the array. At step 5, another “d” has been processed. Since decrementing the frequency for “b” (the item to which the “current” index points) results in a frequency of zero for “b,” the “b” is replaced with the “d,” and the “d” is indicated by a frequency of one.’ Furthermore, the “current” index is moved to point to the next item of the array, containing “c.” At step 6, an additional “b” has been processed. Since the “current” index pointed to “c” with a frequency of one, the “c” is replaced with the “b” and the “b” is indicated by a frequency of one. Furthermore, the “current” index is moved to point to the “a.”
[0028] Thus, at any time, the array includes indications of search term frequencies (perhaps weighted in view of the operation of the algorithm that maintains the indications) for a particular geographic location (or, more generally, for a particular characteristic value on which the events were categorized). FIG. 4 illustrates schematically how the various arrays (denoted by reference numerals 406a1, 406a2, and 406a3; 406b1 and 406b2; 406c1 and 406c2; and 406x1) are “dumped” to a data structure 408 (e.g., a shared memory hash table) accessible for display. In the FIG. 4 example, the data records in the structure 408 each include the query indications and frequency indications (e.g., “a1” may indicate a particular query term and its associated frequency), as well as including an indication of the geographic location the query indications are for and a time corresponding to the record (e.g., the last time a query indication was updated, or the time the data of the record was provided from a partition).
[0029] FIG. 5 illustrates an example of a display 502 that may be generated based on the generated statistics. The portion 504 of the display 502 graphically indicates (e.g., in a map format) geographic locations for indications of query terms and associated frequencies have been determined. In some examples, depending on a level of zoom for which the portion 504 is configured, the locations indicated in the portion 504 may be a function of the frequency of query terms for the locations. In the FIG. 5 example, a user may cause a cursor to hover over the graphic corresponding to a particular location (in the FIG. 5 example, location (a)), which then causes a “bubble” 508 to be displayed providing information about the query terms for that particular location. The information may be, for example, information regarding query terms and corresponding frequency for the particular location. Furthermore, an additional portion 506 may display the query terms for that particular location (e.g., the query terms may be displayed in a scrolling manner in the portion 506).
[0030] FIG. 6 is a flowchart of a method in accordance with an embodiment of the invention, to cause determination and display of the level of popularity of query terms provided to one or more search services. At step 602, event records are received at data collectors from front end servers. At step 604, the event records are provided from the data collectors to a data warehouse. This includes filtering the event records for the search events. At step 606, the filtered search events are partitioned based on geographic location indicated in the search records. At step 608, each partition generates statistics regarding frequency of the search terms for geographic locations of that partition. At step 610, display of generated statistics regarding frequency of the search terms is caused for particular ones of the geographic locations. It is noted that the sequencing shown in FIG. 6 is not meant to imply that such a sequencing is required, as it should be apparent that some of the processing may occur in parallel and/or asynchronously. For example, the steps may occur in “real time” (e.g., in a pipelined fashion) with little or no batching. In fact, in some examples, there may be more buffering than batching.
[0031] We have thus described a method/system to determine frequency of keywords (search query terms) in a search-stream, with a low latency. Additionally, the stream may be partitioned geographically (e.g., by city) and/or temporally (e.g., by hour or by time of day such as morning, afternoon, evening, etc.), allowing a user of the determinations to discern geographic and temporal differences, e.g., in monetizing specific keywords. Search-events from front-end web-servers handling the search queries are processed in real-time. The latency of the frequency determinations may be less than one hour, or even under five minutes. Indications of keyword frequency, may be displayed, for example, relative to a map to allow for easy navigation.
Claims
What is claimed is:
1. A method of generating information indicative of frequency of search terms presented to at least one online search service, comprising:
processing event indications, indicative of user interaction generally with front-end servers such that some of the event indications are indicative of events that are not search events and some of the event indications are indicative of events that are search events, that are being provided from the front-end servers for persistent storage to detect event indications being provided from the front-end servers for persistent storage that are indicative of search events;
persistently storing the event indications that are being provided from the front-end servers and are being processed to detect event indications that are indicative of search events;
processing the detected search event indications of the event indication being provided from the front-end servers for persistent storage and determining based at least in part thereon, by origin location of the events, frequency data indicative of a frequency of each of a plurality of search terms presented to the at least one online search service; and
causing an indication of at least some of the frequency data to be associated with indications of event origin locations to which the frequency data corresponds.
2. The method of claim 1, further comprising:
causing a display to be generated based on the association of the indication of at least some of the frequency data with indications of event origin locations to which the frequency data corresponds.
3. The method of claim 2, wherein:
causing a display to be generated based on the association of the indication of at least some of the frequency data with indications of event origin locations to which the frequency data corresponds includes causing the display to be generated with indications of at least some of the frequency data superimposed on a map based on the indications of event origin locations to which the frequency data corresponds.
4. The method of claim 1, wherein:
processing the detected search event indications of the event indication being provided from the front-end servers for persistent storage and determining based at least in part thereon, by event origin location, frequency data indicative of a frequency of each of a plurality of search terms presented to the at least one online search service includes updating the determination of frequency data based on newly processed detected search event indications of the event indication being provided from the front-end servers for persistent storage.
5. The method of claim 4, wherein:
updating the determinations of frequency data based on newly processed detected search event indications of the event indication being provided from the front-end servers for persistent storage includes applying an algorithm that automatically decays frequency values over time.
Patent Citations (27)
| Patent | Date | Inventor | Cited By |
|---|---|---|---|
| US6202023(B1) | 2001-03-01 | Hancock et al. | Applicant |
| US6757740(B1) | 2004-06-01 | Parekh et al. | Applicant |
| US7051023(B2) | 2006-05-01 | Kapur et al. | Examiner |
| US7146416(B1) | 2006-12-01 | Yoo et al. | Examiner |
| US2003/0004781(A1) | 2003-01-01 | Mallon et al. | Examiner |
| US2004/0199498(A1) | 2004-10-01 | Kapur et al. | Examiner |
| US2004/0225647(A1) | 2004-11-01 | Connelly et al. | Applicant |
| US2004/0249700(A1) | 2004-12-01 | Gross | Examiner |
| US2004/0249713(A1) | 2004-12-01 | Gross | Examiner |
| US2004/0260600(A1) | 2004-12-01 | Gross | Examiner |
| US2004/0260688(A1) | 2004-12-01 | Gross | Examiner |
| US2004/0267604(A1) | 2004-12-01 | Gross | Examiner |
| US2005/0004903(A1) | 2005-01-01 | Tsuda | Applicant |
| US2005/0033803(A1) | 2005-02-01 | Vleet et al. | Applicant |
| US2005/0050215(A1) | 2005-03-01 | Lin et al. | Examiner |
| US2005/0102259(A1) | 2005-05-01 | Kapur | Examiner |
| US2005/0246358(A1) | 2005-11-01 | Gross | Examiner |
| US2005/0246391(A1) | 2005-11-01 | Gross | Examiner |
| US2006/0004704(A1) | 2006-01-01 | Gross | Examiner |
| US2006/0010029(A1) | 2006-01-01 | Gross | Examiner |
| US2006/0074870(A1) | 2006-04-01 | Brill et al. | Examiner |
| US2006/0122994(A1) | 2006-06-01 | Kapur et al. | Examiner |
| US2006/0123014(A1) | 2006-06-01 | Ng | Applicant |
| US2006/0167896(A1) | 2006-07-01 | Kapur et al. | Examiner |
| US2007/0050353(A1) | 2007-03-01 | Ekberg et al. | Examiner |
| US2007/0162379(A1) | 2007-07-01 | Skinner | Examiner |
| US2007/0255701(A1) | 2007-11-01 | Halla et al. | Examiner |
Non-Patent Literature (25)
- Lamm, S.E. and D.A. Reed “Real-Time Geographic Visualization of World Wide Web Traffic”, Computer Networks and ISDN Systems, vol. 28, No. 7, May 1996, pp. 1457-1468.Examiner
- Yahoo! “Yahoo! Launches Yahoo! Buzz Index for Brand Marketers”, press release, Oct. 30, 2000.Examiner
- Chaney, D. “Private Metrics: An Eye to the Keyhole”, MediaPost Publications, downloaded from publications.mediapost.com, Jan. 2001.Examiner
- Pramudiono, I., T. Shintani, K. Takahashi and M. Kitsuregawa “User Behavior Analysis of Location Aware Search Engine”, Proceedings of the 3rd International Conference on Mobile Data Management (MDM'02), Jan. 8-11, 2002.Examiner
- Yahoo! “Yahoo! Buzz Index—World Weekly”, downloaded from buzz.yahoo.com/world, Aug. 9, 2002.Examiner
- Spink, A., S. Ozmutla, H.C. Ozmutla and B.J. Jansen “U.S. Versus European Web Searching Trends”, SIGIR Forum, vol. 36, No. 2, Fall 2002, pp. 32-38.Examiner
- Google “Google Press Center: 2002 Year-End Zeitgeist”, downloaded from www.google/com/press/zeitgeist.html, Feb. 7, 2003.Examiner
- Beitzel, S.M., E.C. Jensen, A. Chowdhury, D. Grossman and O. Freider “Hourly Analysis of a Very Large Topically Categorized Web Query Log”, Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Jul. 25-29, 2004, pp. 321-328.Examiner
- Klemm, A., C. Lindemann, M.K. Vernon and O.P. Waldhorst “Characterizing the Query Behavior in Peer-to-Peer File Sharing Systems”, Proceedings of the 4th ACM SIGCOMM Conference on Internet Measurement (IMC'04), Oct. 25-27, 2004, pp. 55-67.Examiner
- Otsuka, S., M. Toyoda, J. Hira and M. Kitsuregawa “Extracting User Behavior by Web Communities Technology on Global Web Logs”, in Database and Expert Systems Applications, vol. 3180/2004, 2004, pp. 957-968.Examiner
- Google “Google Press Center: Zeitgeist”, downloaded from www.google.com/press/intl-zeitgeist.html, Feb. 8, 2006.Examiner
- Yahoo! “Yahoo! Buzz Index Frequently Asked Questions”, downloaded from buzz.yahoo.com/faq, Mar. 31, 2006.Examiner
- Yahoo! “The Most Searched Term on Yahoo! Canada—Yahoo! Canada”, downloaded from ca.buzz.yahoo.com, Apr. 6, 2006.Examiner
- Saleh, O. and M. Hefeeda “Modeling and Caching of Peer-to-Peer Traffic”, Simon Fraser University, School of Computing Science, Technical Report TR-2006-11, May 2006.Examiner
- Google “Google Trends: About Google Trends”, downloaded from www.google.com/intl/en/trends/about.html, Aug. 10, 2006.Examiner
- Chellapilla, K. and D.M. Chickering “Improving Cloaking Detection Using Search Query Popularity and Monetizability”, Proceedings of the 2nd International Workshop on Adversarial Information Retrieval on the Web, Aug. 10, 2006.Examiner
- Spink, A., H. Partridge and B.J. Jansen “Sexual and Pornographic Web Searching: Trends Analysis”, First Monday, vol. 11, No. 9, Sep. 2006.Examiner
- Stanger, N. “The View from the Chathams: Geovisualization of Web Site Hits using Google Earth™”, Proceedings of the 18th Annual Colloquium of the Spatial Information Research Centre, University of Otago, Dunedin, New Zealand, Nov. 6-7, 2006.Examiner
- Beitzel, S.M., E.C. Jensen, A. Chowdhury, O. Frieder and D. Grossman “Temporal Analysis of a Very Large Topically Categorized Web Query Log”, Journal of the American Society for Information Science and Technology, vol. 58, No. 2, Nov. 22, 2006, pp. 166-178.Examiner
- Pipes, A. “Yahoo! Buzzlist versus Google Trends”, Read/WriteWeb, downloaded from www.readwriteweb.com, Aug. 2, 2007.Examiner
- Google “Google Trends: cathartic”, downloaded from www.google.com/trends, Nov. 8, 2007.Examiner
- Watters and Amoudi, “GeoSearcher: Geospatial ranking of search engine results,” Proceedings of the American Society for Information Science and Technology, vol. 39, Issue 1, pp. 409-416, Published Online: Jan. 31, 2005, http://www3.interscience.wiley.com/cgi-bin/abstract/109882614/.Applicant
- Gish et al., “Geographical Statistics and Characteristics of P2P Query Strings,” School of Electrical Engineering, Tel-Aviv University, Israel, http://research.microsoft.com/workshops/IPTPS2007/papers/shakedGishTankelShavitt.pdf.Applicant
- Lamma et al., “Real-time geographic visualization of World Wide Web traffic,” Source: Computer Networks and ISDN Systems, vol. 28, No. 7, May 1996 , pp. 1457-1468(12). Department of Computer Science, University of Illinois, Urbana, IL, Netscape Communications Corporation, Mountain View, CA, http://www.sciencedirect.com/science?—ob=ArticleURL&—udi=B6TYT-3XY8PKJ-1R&—user=10&—coverDate=05%2F31%.Applicant
- Myllymaki and Kaufman, “High-Performance Spatial Indexing for Location-Based Services,” International World Wide Web Conference archive, Proceedings of the 12th international conference on World Wide Web table of contents, Budapest, Hungary, Session: Information retrieval 2 table of contents, pp. 112-117, Year of Publication: 2003, ISBN:1-58113-680-3.Applicant