KR100574201B1 - A method for detecting seasonal search terms and a system thereof - Google Patents

Abstract

The present invention relates to a method and a system for detecting a search word for each time period in which the number of inputs from a user fluctuates at a specific time, and more specifically, there is a rising period in which the number of inputs from a user exceeds a predetermined reference value. The present invention relates to a method and system for extracting a search term and detecting a search term for each time period corresponding to the rising period by performing a predetermined purification process on the extracted search term.

According to an exemplary embodiment of the present invention, a method for detecting a time-specific search word may include maintaining a database including input number information per predetermined time interval for each search word, analyzing the input number information with reference to the database, and the analysis. And extracting a search word having a predetermined rising period based on the result.

According to the present invention, it is possible to maximize the profits from the sales of the search terms by detecting a search term for each time period in which the popularity increases in accordance with a specific time period, and setting a higher selling price for the detected search term-specific search terms at the next arrival of the specific time period. There is provided a method and system for detecting a time-specific search term.

About seasonal queries, seasonal queries, query sales, ascent periods, input frequency

Description

Search method and system by period {A METHOD FOR DETECTING SEASONAL SEARCH TERMS AND A SYSTEM THEREOF}

1 is a diagram illustrating a network connection of a time-based keyword detection system according to an embodiment of the present invention.

2 is a flowchart illustrating a method of detecting a time period according to an embodiment of the present invention.

3 is a view showing an example of a database in an embodiment of the present invention.

FIG. 4 is a diagram for one example of a graph for analyzing input count information recorded in a database according to one embodiment of the present invention.

5A and 5B illustrate an example of a graph for detecting a search word having a rising period using average and standard deviation according to an embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of a graph for detecting a seasonal search word having a characteristic according to a season according to one embodiment of the present invention.

FIG. 7 is a diagram illustrating an example of a graph for describing a processing procedure for a search word having two or more rising periods according to an embodiment of the present invention.

8 is a block diagram illustrating a system for detecting a time period according to an embodiment of the present invention.

9 is an internal block diagram of a general-purpose computer device that may be employed to perform the time-specific search word detection method according to the present invention.

<Explanation of symbols for the main parts of the drawings>

800: Search term detection system by time

801: Database 802: input frequency analysis means

803: search term extraction means 804: search term re-extraction means

The present invention relates to a method and a system for detecting a search word for each time period in which the number of inputs from a user fluctuates at a specific time, and more specifically, there is a rising period in which the number of inputs from a user exceeds a predetermined reference value. The present invention relates to a method and system for extracting a search term and detecting a search term for each time period corresponding to the rising period by performing a predetermined purification process on the extracted search term.

In general, a search service system that provides a search service includes a search result corresponding to the search word when a user inputs a search word (for example, a web site including the search word, an article including the search word, and a file name including the search word). To the user).

In such a search service system, there may be various search terms input from a user, but in reality, the ratio of the top search terms to the total search requests according to the number of times input from the user is very high, and the ratio of the remaining search terms to the total search requests is Relatively low.

This can be done by setting a higher selling price for the top search terms with a high number of search requests, thereby maximizing the profits from the sale of the search terms, and at the same time, a user through a so-called customization service that provides search results intensively for the top search terms. It means you can provide a higher level search service.

On the other hand, if you analyze the top search terms, 1) there is a search term that is always ranked at the top (for example, "game", "go-stop", "celebrity", etc.) 2) a certain time of year Accordingly, there is a case where the ranking rises rapidly and belongs to a higher search word (for example, "carnation", "swimsuit", "rice cake soup", etc.).

In the case of the search term corresponding to 1), the selling price is always higher than the selling price for the lower search terms having a low number of search requests from the user, and the user is continuously managed to the higher level to provide the user with optimized information. Needs to be.

However, in the case of the search term corresponding to 2), the popularity increases sharply in accordance with the arrival of a certain period of time, but when the specific period ends, the popularity falls sharply and reaches a congestion state. It is not possible to continuously invest time and economic efforts.

For example, "carnation" is rapidly increasing the number of inputs from the user from April 25, in line with the arrival of May, which is Mother's Day and Teacher's Day. It may be a falling search term.

In the case of such a query, if the sales unit price is set high for a period other than April 25 to May 15, it is natural that the sales of the query will not be performed smoothly. Continued efforts to tailor services over time can result in significant time and economic losses for the search engine operator who runs the search engine.

Therefore, it is a very efficient way to increase the sales price of the search term "carnation" and make a big profit only for the period from April 25 to May 15, and to lower the sales price again for other periods. May also be available only between April 25th and May 15th, to provide users with a personalized service for the term "carnation" to satisfy their desire to receive optimized search results, In terms of, managing the search word to a general degree may be a very reasonable method for managing the search word.

On the other hand, a general search term such as "carnation" may be predicted that the popularity will surge in May even with a guess of a person, but there may be a search term that is rapidly increasing in popularity at a certain time without being predictable. For example, it is not easy to predict when the search term "novel book" will surge in popularity, but after analyzing the number of inputs from the user, the popularity surged around 15 January, and again around 30 January. You can find it falling. When the facts are used to sell search words, advertisements, and the provision of personalized services, the profits of search engine operators may be maximized, and user needs may be satisfied.

In addition, even in the case of a search word that is expected to decrease to a certain extent after a sudden increase in popularity with the arrival of a certain time, it is essential to clearly and accurately specify the time period for selling the search word and providing customized services.

However, search service providers are currently unable to detect and detect so-called spiked search terms, in which the number of inputs from users rises and falls again. There is no method available.

Accordingly, there has been a demand for the emergence of a new technology capable of continuously observing and analyzing the number of inputs from a user for a search word, and thus detecting a search word having a characteristic according to a time to be used for selling a search word and providing a customized service.

The present invention has been made to solve the problems of the prior art as described above, detects a time-specific search terms that increase in popularity in accordance with a specific time, sales for the detected time-specific search terms when the next arrival of the specific time It is an object of the present invention to provide a method and a system for detecting a time-specific search word that can be set to a high unit price to maximize the profit from the search word sales.                         

In addition, the method and system for detecting time-phased search terms according to the present invention detects time-phased search terms that increase in popularity in accordance with a specific time period, and through intensive advertisements for the detected time-phased search terms around the coming of the specific time period. Its purpose is to enable active search word sales.

In addition, the method and system for detecting time-specific search terms according to the present invention detects time-specific search terms that increase in popularity according to a specific time period, and when the next arrival of the specific time period, the user searches for optimized search results for the detected time-specific search terms. In order to satisfy the user's desire to be provided with a personalized service for a popular search word at a specific time.

In addition, the method and system for detecting time-phased search terms according to the present invention detects a search term that cannot be easily predicted by the time-specific search term as a time-specific search term by synthesizing and analyzing data accumulated in all fields for a certain period of time. For that purpose.

In addition, the method and system for detecting time-phased search terms according to the present invention can be utilized for the sales of search terms, advertisements, and customized services by precisely specifying the point of time after the rapid rise and fall of popularity by analyzing and analyzing data accumulated in all fields for a certain period of time. Its purpose is to

In order to achieve the above object and to solve the above-mentioned problems of the prior art, the time-based search term detection method according to an embodiment of the present invention includes information on the number of times input per predetermined time interval for the search word Maintaining a database, analyzing the input count information with reference to the database, and extracting a search word having a predetermined rising period based on the analysis result.

According to an aspect of the present invention, the method may further include re-extracting a search word that satisfies a predetermined re-extraction criterion among the extracted search terms, wherein the re-extraction criterion includes a first reference value and a second predetermined length of the rising period. There is provided a time-based search term detection method characterized by being associated with whether there is a reference value.

According to another aspect of the present invention, the method further includes re-extracting a search word satisfying a predetermined re-extraction criterion among the extracted search terms, wherein the database includes the input number of times information for a predetermined period, Generating first hit number information by summing input count information for a period, generating second hit count information by summing input count information for the remaining period except the rising period in the predetermined period, and Dividing the 1 hit number information by the second hit number information to calculate a share corresponding to the rising period, wherein the re-extraction criterion is related to whether the share exceeds a predetermined reference value. There is provided a method for detecting a search word according to time.

In accordance with another embodiment of the present invention, a time-based search term detection system includes a database including input count information per predetermined time interval for the search word for each search word, and an input count analysis number for analyzing the input count information by referring to the database. However, and a search word extracting means for extracting a search word for a predetermined rise period based on the analysis result.

Hereinafter, with reference to the accompanying drawings will be described in detail the search method and system for each time according to the present invention.

1 is a diagram illustrating a network connection of a time-based keyword detection system according to an embodiment of the present invention. Users access the time-based search term detection system 100 through a wired or wireless communication network using the user terminal 110a or 110b and input a search term. The search term detection system 100 for each time records the number of inputs for the search term in a predetermined database, and extracts the search term having a rising period by analyzing the input number. In addition, the time-based search term detection system 100 according to another embodiment of the present invention re-extracts a search term that satisfies a predetermined re-extraction criterion among the extracted search terms, or excludes some search terms according to a predetermined exclusion criterion. In this way, a search term for each time corresponding to the rising period may be detected.

The time-based search term detection system 100 according to the present invention may be operated in conjunction with an Internet search service system, so that a user accesses an Internet search service system in the ascending period, which comes next, to search for the detected time-related search term. If entered, the user may be provided with optimized search information for the search word. In addition, the time-based search term detection system 100 according to the present invention may be operated in conjunction with a keyword sales system for an internet search service, and thus, in the ascending period coming next to a number of advertisers who want to purchase keywords. In accordance with the above, the price of the detected search term for each time is set high and the profit of the Internet search service provider can be maximized.

2 is a flowchart illustrating a method of detecting a time period according to an embodiment of the present invention. The keyword search method for each time period according to the present embodiment is provided in a predetermined search term detection system.

In step 201, the time-based search term detection system maintains a database including input count information per predetermined time interval for the search term for each search term.

3 is a diagram illustrating an example of the database. In FIG. 3, a search term identifier 301 is used as a means for identifying each search term. As shown in FIG. 3, the database may include input count information per predetermined time interval for a search word. Referring to the reference numeral 302 shown in FIG. 3, the database has input number information "500" and "400" per predetermined time interval for "parental day" with the search term identifier "A54095" associated with "parental day". "," 50 "and the like are recorded. Although FIG. 3 shows an example of a database in which the time interval is "day", the time interval is set by the operator of the time-based search term detection system as "day", "two days", "one week", "12 hours", or the like. Various determinations may be made and the operator of the system may change an existing time interval to another time interval.

According to an embodiment of the present invention, the daily input count information may be accumulated and stored for each search word requested by users for one year in the database. In the present embodiment, a search term for each time having monthly or seasonal characteristics may be detected through an analysis and extraction process to be described below.

In step 202, the time-specific search word detection system analyzes the input frequency information by referring to the database.

According to an embodiment of the present invention, the time-based search term detection system shows a graph according to the input count information recorded in the database in step 202, and analyzes the variation of the input count information using the graph. I can do it. 4 shows an example of such a graph. As illustrated in FIG. 4, daily input frequency information for one year for the search term "double eyelid" is shown in the form of a graph.

In FIG. 4, the straight lines parallel to the horizontal axis mean reference numerical information for size comparison with the daily input frequency information. The reference numerical information may be determined by various methods, which will be described later.

In the present invention, a period in which the number of times input for a search word exceeds this reference numerical information is defined as an "rising period." Therefore, this rising period may appear in the case of a search word that tends to increase rapidly as the number of inputs from the user increases with the arrival of a specific time and then decreases to a general number of inputs after a certain time. As such, according to one embodiment of the present invention, the rising period may be associated with a change in the input count information recorded in the database.

For example, when the straight line corresponding to the reference numeral 403 is determined as reference numerical information in FIG. 4, the search term “double eyelid” may indicate that the number of times of input of the reference numeral information corresponds to the reference numeral 401. Since the input count information falls below the reference numerical information at the time corresponding to the reference 402, the period from the time corresponding to the reference 401 to the time corresponding to the reference 402 is determined. It can be defined as the rising period of the search term "double eyelid". As such, the rising period may be a period from the time when the input count information exceeds the predetermined rising start reference value to the time when the falling number falls below the predetermined rising end reference value.

On the other hand, in the above example, the rising start reference value and the rising end reference value are both the same as the linear values corresponding to the reference numeral 403, but according to an embodiment of the present invention, the rising start reference value and the rising end reference value may be different from each other. It may be.

For example, in FIG. 4, a straight line corresponding to reference numeral 404 is determined as reference numerical information for an ascending start reference value, and a straight line corresponding to reference numeral 403 is determined as reference numeric information for an ascending end reference value. The search term " double eyelid " indicates that the number of times of input information exceeds the starting point of reference value at the time corresponding to the reference numeral 405, and the number of times of input information decreases below the rising end reference value at the time of reference numeral 402. Therefore, the period from the time point corresponding to the reference numeral 405 to the time point corresponding to the reference numeral 402 may be defined as the rising period of the search term "double eyelid".

According to an embodiment of the present invention, there is provided a time-specific search word detection method, characterized in that the rising end reference value is set lower than the rising start reference value. If the number of inputs to a search term is fluctuating, it may happen that a short time after the start of the start of the rise above the start end threshold, it is determined that there is no rise period or a very short period of time. It may be determined that the problem is not detected as a time-specific search word. However, according to this embodiment, the above problem can be prevented by setting the rising end reference value lower than the rising start reference value.

Meanwhile, as described above, the reference numerical information for determining the rising end reference value and the rising end reference value may be determined by various methods. According to an embodiment of the present invention, the time-based search term detection system may include the reference numerical information. In order to generate, the average value information and the standard deviation information regarding the input count information for a predetermined period may be generated with reference to the database, and the reference numerical information may be generated using the average value information and the standard deviation information.

In this case, the reference numerical information may be generated by adding the average value information and a value obtained by multiplying the standard deviation information by a predetermined number. For example, when the predetermined number is 1, the reference numerical information may be determined as "average value + (1 * standard deviation)" of the number of inputs. As another example, in the embodiment in which the rising end reference value is set lower than the rising start reference value as described above, the rising start reference value is determined as “average value + (3 * standard deviation)” and the rising end reference value is “average value + (1). Standard deviation).

5A and 5B are diagrams showing an example of a graph for detecting a search word in which a rise period exists using average and standard deviation as in the present embodiment.

FIG. 5A shows a graph showing the number of input information for a year for a search term of "Christmas" in the form of a graph, and reference numeral 503 adds the standard deviation information to the average value information for the number of input information for one year. The generated reference numerical information is shown in a straight line. As shown in Fig. 5A, the search term " Christmas " has the number of times of input at the time corresponding to the reference numeral 501 above the reference numerical information, and the number of times of input at the time corresponding to the reference numeral 502. Since the information falls below the reference numerical information, the period from the point corresponding to the reference numeral 501 to the point corresponding to the reference numeral 502 may be defined as the rising period of the search term "Christmas".

FIG. 5B shows the number of input information for 30 days for the search term "parent's day" in the form of a graph, with reference numeral 511 representing the number of input information, and reference numeral 512 for 30 days. The reference numerical information generated using the average value information and the standard deviation information regarding the input count information is expressed. As shown in FIG. 5B, the search term "parental day" has the input count information at the time corresponding to the reference numeral 513 above the reference numerical information, and the input at the time corresponding to the reference numeral 514. Since the retrieval information falls below the reference numerical information, the period from the time corresponding to the reference numeral 513 to the time corresponding to the reference numeral 514 may be defined as a rising period of the search term "parent's day".

In addition, according to another embodiment of the present invention, the time-based search term detection system refers to the database, by using the EWMA (Exponentially Weighted Moving Average) management average value information about the input frequency information for a predetermined period and Generating management upper limit information, and determining whether the average value information is moved based on the input count information and the management upper limit information, and the ascending period is a period during which the average value information is determined to have been moved. A search word detection method is provided.

In general, control charts are widely used for various purposes, but among them, EWMA charts are most commonly used because they can be easily generalized and small changes are easily detected. In the manufacturing process, it is very important to continuously monitor the quality of the product during the production process in order to produce the product quality to the standard. EWMA control chart is generally used to solve this problem.

According to the present embodiment, by using the EWMA control chart to detect the search term for each time, it is possible to monitor the trend of the number of inputs according to the time change for any one of the search terms.

, 분산이

인 정규분포를 따른다고 하고, 본 실시예에 따른 시기별 검색어 검출 시스템은 매일 상기 검색어에 대한 평균의 이동 여부를 판단한다. 상기 시기별 검색어 검출 시스템이 특정 시점에서 상기 평균이 이동되었다고 판단하는 것은 사용자들이 상기 검색어에 대한 관심이 이전과 다르다는 것을 의미하고, 상기 시기별 검색어 검출 시스템은 상기 평균치 정보가 이동된 것으로 판단된 기간을 상기 검색어에 대한 상승 기간으로 판단할 수 있다.The search term detection method according to the present embodiment may be made by the following statistical problem. First of all, the average number of times

, Dispersion

A normal search distribution according to the present invention, and the time-based search term detection system according to the present embodiment determines whether the average for the search word moves daily. The determination that the mean has been shifted at a specific point in time by the search term detection system means that users are less interested in the search word than before, and the search term detection system by time is determined that the mean information has been moved. May be determined as a rising period for the search word.

This problem is solved using the EWMA chart. Considering the following statistic at time t,

는 t시점에서 측정된 입력 회수

를 p의 가중치를 두어 고려하고 바로 전 시점에서의 통계량에 대해서는 (1-p)의 가중치를 두어 계산한다. 이를 통해서, 최근 측정치에 보다 많은 비중을 두어 평균 이동을 고려할 수 있게 된다. 수식 1은 다음과 같이 다시 표현할 수 있다.statistic

Is the number of inputs measured at time t.

Calculate the weighting factor of p and calculate the weighting factor of (1-p) for the statistics just before. Through this, the average shift can be considered by giving more weight to recent measurements. Equation 1 can be rewritten as follows.

의 분산은 다음과 같이 계산될 수 있다.Meanwhile, statistics

The variance of can be calculated as

In addition, using the general control chart theory, the following management upper limit can be calculated.

는

을 불편추정량으로 만들어 주는 계수로 1.128을 사용할 수 있고,

은 이동 범위를 평균한 수치이다. 즉,

표현할 수 있다.In Equation 4,

Is

You can use 1.128 as a factor to make an uncomfortable estimate,

Is the average of the moving ranges. In other words,

I can express it.

가 UCL을 넘어간다면 평균이 이동했다고 판단할 수 있고, 이렇게 판단된 검색어를 시기별 검색어로 검출할 수 있다. According to the method for detecting a time-specific search word using the EWMA control chart according to the present embodiment, the time-based search word detection system is calculated every time

If the average exceeds the UCL, it may be determined that the mean has shifted, and the determined search term may be detected as a time-specific search term.

, 현재 자료를

이라 하고, 2) 이전 90일 자료를 이용해서 EWMA 관리도의 평균과 관리상한선을 계산한 후, 3) 최종 관리도가 만들어지면, 현재 자료

을

으로 변환하여 상기 관리도에서

이 관리상한선을 상회하는지 여부를 확인하여 평균의 이동여부를 판단할 수 있다. 여기서, 상기와 같이 90일의 자료를 고려하는 것은 하나의 실시예에 불과하며, 30일, 40일, 60일 등과 같이 여러 기간 동안의 자료를 고려할 수 있음은 당연하다.As an example of a method of detecting a search term based on the current data and the previous 90 days, the search term detection system according to the time period may include 1) the previous 90 days data

Current data

2) Calculate the average and upper limit of the EWMA chart using the previous 90 days of data.

of

To convert from

It is possible to determine whether or not the average shifts by checking whether the management upper limit is exceeded. Here, considering the data of 90 days as described above is only one embodiment, it is natural that the data for various periods, such as 30 days, 40 days, 60 days can be considered.

In step 203, the time-based search term detection system extracts a search term having a rising period based on the analysis result in step 202. For example, in the case of FIG. 4, the search term "double eyelid" has a rising period from the time point corresponding to the reference numeral 401 or the reference numeral 405 to the time point corresponding to the reference numeral 402. ), The time-based search term detection system may extract a search term "double eyelid". In addition, the time-specific search word detection system may detect the search word extracted in step 203 as a time-specific search word corresponding to the rising period.

On the other hand, according to an embodiment of the present invention, in step 204, characterized in that the time-based search term detection system to extract a search word that satisfies a predetermined re-extraction criteria among the search words extracted in step 203 again According to the present invention, a method for detecting a specific keyword is described below. According to this embodiment, it is possible to detect a search term for each time as a more refined result.

According to the present embodiment, the re-extraction criterion is related to whether the length of the rising period is present between a predetermined first reference value and a second reference value, and wherein the first reference value and the second reference value are the time-specific search terms to be detected. It can be determined according to the type of.

For example, in order to detect a monthly search word having a monthly characteristic, the first reference value may be determined as 5 days, and the second reference value may be determined as 30 days, and an increase period exists but the length does not exist between 5 and 30 days. The search term may not be extracted in step 204 because the search term does not satisfy the re-extraction criteria.

As another example, in order to detect seasonal search terms having seasonal characteristics, the first reference value may be determined as 40 days, and the second reference value may be determined as 120 days, and an increase period exists but the length does not exist between 40 and 120 days. The search term may not be extracted in step 204 because the search term does not satisfy the re-extraction criteria.

6 is a diagram illustrating an example of a graph for detecting the seasonal search word as described above. In Fig. 6, the number of inputs for a year for the search term "spring clothes" is shown in the form of a graph, and reference numeral 601 is shown as reference numerical information. In addition, it can be seen that reference numeral 602 corresponds to January 28, 2004 as the start of the ascent, and reference numeral 603 corresponds to April 1, 2004 as the end of the ascension. The term "spring clothes" rises from January 28, 2004 to April 1, 2004, so the length is 63 days. As in the previous example, the first reference value is 40 days and the second reference value is 120 days. In this case, the search term "spring clothes" satisfies the re-extraction criteria, and thus may be extracted as a seasonal search term in step 204.

According to an embodiment of the present invention, in the case of a search word having two or more rising periods, the search term detection system for each period determines whether the interval between the first rising period and the second rising period is equal to or less than a predetermined reference interval. When the interval between the first and second rising periods is equal to or less than the reference interval, the first and second rising periods and the intervals are added to generate one third rising period, so that the length of the third rising period is In accordance with a presence or absence between the first reference value and the second reference value, a search term-specific search method is provided in step 204. At this time, the reference interval can be appropriately determined experimentally, it is possible to vary.

7 is a diagram illustrating an example of a graph for explaining a processing procedure for a search word having more than one rising period according to the present embodiment. In FIG. 7, the number of times of daily input for a search word is denoted by reference numeral 701, and reference numeral information is denoted by reference numeral 702, respectively.

Referring to FIG. 7, the daily input frequency information for this search word starts to exceed the standard numerical information on May 12, and falls below the standard numerical information on May 14, and the standard numerical information on May 15 again. It is likely to start to outperform and fall below baseline data on 17 May.

In this case, the period from May 12 to May 14 is the first ascent period, and the period from May 15 to May 17 is the second ascent period, from May 14 to May 15 The periods up to can be defined at intervals of both rising periods. If the first and second rise periods are treated as separate rise periods due to the existence of the interval between the two rise periods, the rise period for the search term is only two days, but both rise periods and the intervals are combined. In terms of the rising period of, the rising period for this search term is 5 days. The time-based search term detection system according to the present embodiment views the rising period for the search term as 5 days instead of 2 days, and according to whether or not the length of the rising period of 5 days exists between the first reference value and the second reference value ( At 204, the search term may be re-extracted.

According to the present exemplary embodiment, a search term determined to be relatively high in the number of inputs over a short period of time, but may be detected as a term search term, may be excluded from the term search term on a uniform basis. All the search terms that can be detected as a search term by time.

Meanwhile, according to an embodiment of the present invention, the database maintained in step 201 includes the input count information for a predetermined period, and the search term detection system for each time period is configured to display the input count information for the rising period. Summing up to generate first number of views information, sum the number of times of input information for the remaining period other than the rising period in the predetermined period to generate second number of views information, and the first number of views information to the second number of views information There is provided a method for detecting a time-specific search word, which divides to calculate a share corresponding to the rising period, and re-extracts a search word whose share exceeds a predetermined reference value in step 204.

In practice, in order to detect meaningful time-phased search terms, the number of inputs over a certain time period needs to be considerably compared with the number of inputs over a period except that time period, and the present embodiment reflects this need. According to the present embodiment, a search term for each time as a more refined result can be detected.

In addition, according to an embodiment of the present invention, in the case of a search word having two or more rising periods, the search term detection system for each time period adds the number of times of input for the first rising period in calculating the occupancy rate. The first hit number information may be generated, and the second hit number information may be generated by adding up the number of times of input for the remaining period except the first rising period and the second rising period in the predetermined period. The search term detection system for each time period divides the first view count information generated as described above into second view count information to calculate a share corresponding to the first rising period, and searches for a search word whose share exceeds a predetermined reference value. Re-extraction at step 204.

As an example according to the present embodiment, when the search term "Chosun Hotel" has a rise period of January and August of the year, when the occupancy rate corresponding to January is calculated, information on the number of inputs during August is displayed. The first hit information is generated by adding the first count information, and the second hit count information is generated by summing the input count information for the remaining periods except for 1 year to January as well as August, and the first hit information and the second hit number generated as described above are generated. The information is used to calculate the January share of the search term "Chosun Hotel."

In the case of a search term having more than one rise period, there may be a problem in that the share is low because the number of inputs in another rise period is too high, and thus it is not detected as a search term for each time period. Will be prevented.

Meanwhile, according to an embodiment of the present invention, the search term detection system for each time period generates average value information regarding the number of times of input information during an ascending period, and in step 204, the search term for which the average value information exceeds a predetermined reference value. Provided is a method for detecting a time-specific search word, which is characterized by re-extracting from.

Despite the fact that the number of inputs during the ascending period is a very low search term, the problem can be detected as a time-based search term simply because it exceeds the standard number. The number of times needs to exceed a certain number, and this embodiment reflects this need. According to the present embodiment, only a search word whose average number of inputs records a predetermined value or more is detected as a time-specific search word. Can be prevented in advance.

Meanwhile, according to an embodiment of the present invention, in step 205, the time-based search term detection system excludes some search terms from the search terms extracted in step 203 or 204 according to a predetermined exclusion criterion. There is provided a method of detecting a search word according to time. Here, the exclusion criterion may be related to whether the search term is included in a predetermined name dictionary or whether the search term is included in a predetermined social issue search term dictionary that stores a search term associated with a social issue.

The name dictionary is a kind of storage means including the name of a person, and the term-based search term detection system according to the embodiment of the present invention searches a term extracted by referring to the term dictionary to prevent the name of the person from being detected as the term-specific search term. You can exclude a person's name from it. In addition, the social issue search term dictionary may be another storage means including a search word produced and managed as social issue content by an operator of a search engine due to being a social issue. The system may exclude a search word associated with a social issue from a search word extracted by referring to the social issue search term dictionary to prevent a search word that appears according to a social issue as a search term for each time. As described above, according to the present exemplary embodiment, it is possible to not detect a search word as a time-specific search word, in which the number of inputs is recorded by accident at that time regardless of time-specific characteristics.

Meanwhile, according to an embodiment of the present invention, in step 206, the time-based search term detection system may include steps 201 to 203, or in addition to steps 204 and 205, for a predetermined time. Periodically, each time period, characterized in that the search term is detected as a time-specific search term over a predetermined number of times in the execution result step 207 as a final time-specific search term corresponding to the rising period A detection method is provided. If the time period is one year, in step 206, the time-based search term detection system performs steps 201 to 205 every year to detect year-by-year search terms, and in step 207 At the same time, a search word that records a rising period may be finally detected as a search word for each time period.

For example, even though the term "novel book" was detected as a time-based search term that recorded a month-long rise in May, detecting "novel book" as a monthly search term that increases in popularity every May each year There may be problems that may be somewhat inaccurate. However, if the term "novel book" is detected as a time-based search term that rose in the month of May, either in the next year or the following year, the search term "novel book" may be a monthly search term that increases in popularity every May. Since this is so high, detecting it as a monthly search word may correspond to a very accurate result.

As described above, according to the present embodiment, by searching and analyzing data accumulated over a long period of time as a more accurate result, a search term that cannot be easily predicted by a specific time period as a search term by time is used as a search term. The detection effect can be obtained.

Further, according to an embodiment of the present invention, the predetermined search term is detected as a time-specific search term corresponding to the first rising period in the first time period and the time-specific search term corresponding to the second rising period in the second time period. If the difference between the first and the second rise period is within a predetermined error range, the search term-specific search term detection system determines that the search term is a time period corresponding to one third rise period and the third rise period. Is determined based on at least one of the first and second rising periods and the difference.

For example, the search term "piano" recorded the first up period from March 1 to March 15 in 2003, and the second up period from March 5 to March 17 in 2004. In the case of recording, the time-based search term detection system according to the present exemplary embodiment analyzes the difference between the first rise time and the second rise period and determines that the search term “piano” is one rise period ( For example, the keyword may be detected by a time-specific search word corresponding to March 1 to March 17). Since it is very unlikely that the number of inputs to a search term accurately records the rise period at a certain time every year, according to the present embodiment, if the difference between both rise periods is within a certain error range, It can be detected by the search term for each recording time. As a result, it is possible to obtain an effect of accurately detecting all the time-specific search terms that can be excluded from the detection by a uniform criterion.

In addition, embodiments of the present invention include computer-readable media containing program instructions for performing various computer-implemented operations. The program recorded on the computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. The media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts.

Hereinafter, a search term detection system for each time period according to an embodiment of the present invention will be described. 8 is a block diagram illustrating a system for detecting a time-specific keyword according to an exemplary embodiment.

The time-based search term detection system 800 according to the present embodiment includes a database 801, an input frequency analysis means 802, and a search term extraction means 803.

The database 801 includes information on the number of times of input per predetermined time interval for the search word for each search word.

3 is a diagram illustrating an example of the database 801. Since FIG. 3 has already been described herein, a detailed description thereof will be omitted. Meanwhile, according to an embodiment of the present invention, the daily input count information may be accumulated and stored in the database 801 for each search word requested by users for one year. In the present embodiment, a search term for each time having monthly or seasonal characteristics may be detected through an analysis and extraction process to be described below.

The input count analyzing means 802 analyzes the input count information with reference to the database 801, and the search word extracting means 803 extracts a search word having a rising period based on the analysis result.

According to an embodiment of the present invention, the input frequency analysis means 802 may show a graph according to the input frequency information recorded in the database 801, and analyze the variation of the input frequency information using the graph. have. 4 shows an example of such a graph. Since FIG. 4 has already been described herein, a detailed description thereof will be omitted.

According to one embodiment of the invention, the rise period may be associated with a change in the input count information recorded in the database 801, with the result that the rise period rapidly rises as the number of inputs from the user arrives at a certain time. This may occur in the case of a search term that tends to fall back to a general number of inputs after a certain time. Further, according to another embodiment of the present invention, the rising period may be a period from the time when the number of times of input information exceeds a predetermined rising start reference value to a time of falling below a predetermined rising end reference value, and the rising start reference value and the rising time. The termination reference values may be the same or may be different from each other.

Meanwhile, according to an embodiment of the present invention, the time-based search term detection system 800 further includes a search term re-extraction means 804, and the search term re-extraction means 804 is a search term extracted by the search term extraction means 803. The search word that satisfies the predetermined re-extraction criteria may be extracted again. According to this embodiment, it is possible to detect a search term for each time as a more refined result. In particular, the search term re-extracting means 804 includes a first reference value and a second reference value having a predetermined length of the rising period among the extracted search terms. You can re-extract the search terms that exist between them.

The first reference value and the second reference value may be determined according to the type of search term for each time to be detected. For example, in order to detect a monthly search word having monthly characteristics, the first reference value may be determined as 5 days, and the second reference value may be determined as 30 days. The search term re-extraction means 804 may have a rising period, but the length may be increased. Search terms that do not exist between 5 and 30 days may not be extracted because they do not satisfy the re-extraction criteria.

As another example, in order to detect seasonal search terms having seasonal characteristics, the first reference value may be determined as 40 days, and the second reference value may be determined as 120 days, and the search term re-extraction means 804 may have a rising period, but Search terms that do not exist between 40 and 120 days may not be extracted because they do not satisfy the re-extraction criteria. 6 is a diagram illustrating an example of a graph for detecting the seasonal search word as described above. Since FIG. 6 has already been described herein, a detailed description thereof will be omitted.

Meanwhile, according to an embodiment of the present invention, the database 801 includes the input frequency information for a predetermined period, and the time-based search term detection system 800 includes the first hit number information generating means 805 and the second. The hit number information generating means 806 and the occupancy rate calculating means 807 may further be included.

The first hit count information generating means 805 generates the first hit count information by adding up the number of times of input information during the rising period, and the second hit count information generating means 806 performs the remaining period except the rising period in the predetermined period. The second number of views information can be generated by summing up the input number of times information.

The occupancy calculating means 807 divides the first hit number information by the second hit number information to calculate a occupancy rate for calculating the occupancy corresponding to the rising period, and the search term re-extraction means 804 according to the present embodiment. May re-extract a search word of the search word extracted by the search word extracting unit 803 that exceeds the predetermined reference value.

In practice, in order to detect meaningful time-phased search terms, the number of inputs over a certain time period needs to be considerably compared with the number of inputs over a period except that time period, and the present embodiment reflects this need. According to the present embodiment, a search term for each time as a more refined result can be detected.

9 is an internal block diagram of a general-purpose computer device that may be employed to perform the time-specific search word detection method according to the present invention.

The computer device includes one or more processors 910 coupled with a main memory device including random access memory (RAM) 920 and read only memory (ROM) 930. The processor 910 is also called a central processing unit (CPU). As is well known in the art, the ROM 930 serves to transfer data and instructions to the CPU unidirectionally, and the RAM 920 typically transfers data and instructions bidirectionally. Used to. RAM 920 and ROM 930 may include any suitable form of computer readable media. Mass storage 940 is bidirectionally coupled to processor 910 to provide additional data storage capability and may be any of the computer readable recording media described above. The mass storage device 940 is used to store programs, data, and the like, and is a secondary memory device such as a hard disk which is generally slower than the main memory device. Certain mass storage devices such as CD ROM 960 may be used. The processor 910 may include one or more input / output interfaces, such as video monitors, trackballs, mice, keyboards, microphones, touchscreen displays, card readers, magnetic or paper tape readers, voice or handwriting readers, joysticks, or other known computer input / output devices. 950. Finally, the processor 910 may be connected to a wired or wireless communication network through the network interface 970. Through this network connection, the procedure of the method described above can be performed. The apparatus and tools described above are well known to those skilled in the computer hardware and software arts.

The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention.

As described above, although the present invention has been described with reference to limited embodiments and drawings, the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined not only by the claims below, but also by the equivalents of the claims.

According to the method and system for detecting time-specific search terms according to the present invention, a time-specific search term that increases in popularity according to a specific time is detected, and when the next arrival of the specific time is set, the unit price for the detected time-related search term is set high. It is possible to maximize the profits from the search word sales.

In addition, according to the method and system for detecting time-specific search terms according to the present invention, a time-specific search term of increasing popularity in accordance with a specific time period is detected, and an intensive advertisement for the detected time-specific search term is about the next arrival of the specific time period. Through the active query sales can be achieved.

In addition, according to the method and system for detecting time-specific search terms according to the present invention, a time-specific search term that increases in popularity according to a specific time period is detected, and an optimized search result for the detected time-specific search terms when the next arrival of the specific time period occurs. By providing to the user, it is possible to obtain the effect of satisfying the user's desire to be provided with a personalized service for a popular search term at a specific time.

In addition, according to the method and system for detecting time-specific search terms according to the present invention, by searching and analyzing the data accumulated in all fields for a certain period of time as a search term that cannot easily predict the increase in popularity according to a specific time period. You can get the effect.

In addition, according to the method and system for detecting time-specific search terms according to the present invention, by accumulating and analyzing data accumulated in all fields for a certain period of time, precisely specifying a time point after a sudden increase in popularity and providing a search service, an advertisement, and providing a customized service. The effect can be obtained.

Claims (21)

delete delete In the method for detecting a time-specific search term by the predetermined time-term search word detection system that fluctuates the number of inputs from the user at a specific time, Maintaining, by the database of the time-based search word detection system, including input count information for a predetermined time interval for the search word for each search word; Analyzing, by the input frequency analysis means of the search term detection system for each time, the input frequency information for each keyword by reference to the database; Extracting, by the search term extracting means of the search term detection system for each time period, a search term having a rise period corresponding to a change in the input count information based on the result of the analysis; A time period from a time point of exceeding to a time point of falling below a predetermined rising end reference value, wherein the rising start reference value and the rising end reference value are determined for each search word; And Re-extracting a search word that satisfies a predetermined re-extraction criterion among the extracted search terms, by the search word re-extraction means of the search term detection system by time; For search terms that have more than one of these elevation periods, Determining, by the search word re-extraction means, whether an interval between a first rising period and a second rising period is equal to or less than a predetermined reference interval; And If the interval between the first and second rising periods is equal to or less than the reference interval, as a result of the determination, generating the third rising period by adding up the first, second rising periods, and the intervals; The re-extraction criterion is related to whether the length of the third rising period exists between the first reference value and the second reference value, and wherein the re-extracted search word is detected as a time-specific search word corresponding to the rising period. Characteristic search method for each time period characterized by. The method of claim 3, Generating, by the input count analyzing means, the average value information and the standard deviation information regarding the input count information for a predetermined period by referring to the database; And Generating reference numerical information using the average value information and the standard deviation information; More, And the rising start reference value and the rising end reference value are determined according to the reference numerical information. The method of claim 4, wherein The step of generating reference numerical information using the average value information and the standard deviation information, And generating the reference numerical information by summing a value obtained by multiplying the standard deviation information by a predetermined number and the average value information. In the method for detecting a time-specific search term by the predetermined time-term search word detection system that fluctuates the number of inputs from the user at a specific time, Maintaining, by the database of the time-based search word detection system, including input count information for a predetermined time interval for the search word for each search word; Generating, by the input frequency analysis means of the search term detection system for each time period, the average value information and the upper limit management information about the input frequency information for a predetermined period of time by using the EWMA (Exponentially Weighted Moving Average) chart; Determining, by the input count analyzing means, whether the average value information is moved based on the input count information and the upper limit management information; And Extracting, by the search term extracting means of the search term detection system, a search term having a rise period based on the determination result, wherein the rise period is determined as a period in which the average value information is determined to be moved; Including, The search term for each time period characterized in that the extracted search term is detected as a time-specific search term corresponding to the rising period. delete delete The method of claim 3, And the first reference value and the second reference value are determined according to the type of search term for each time to be detected. The method of claim 9, The search term for each time period, characterized in that the type of the search term for each time period includes a seasonal search term or a monthly search term. delete The method of claim 3, The database includes the input count information for a predetermined period of time, Generating, by the first inquiry number information generating means of the search term detection system for each period, the first inquiry number information by adding up the number of times of the input information during the rising period; Generating, by the second inquiry count information generating means of the search term detection system for each period, the second count information by adding up the number of times of input for the remaining period except the rising period in the predetermined period; And Occupancy calculating means of the search term detection system for each time period divides the first hit number information by the second hit number information to calculate an occupancy rate corresponding to the rising period More, And wherein the re-extraction criterion is further related to whether the occupancy exceeds a predetermined reference value. The method of claim 12, For search terms that have more than one of these elevation periods, The adding of the number of times of input information during the ascending period to generate the first inquiry number information may include: Generating the first inquiry number information by adding up the number of times of input information during a first rising period; The step of generating the second inquiry information by summing the input count information for the remaining period except the rising period in the predetermined period, Generating the second inquiry number information by adding up the input count information for the remaining period except the first rising period and the second rising period in the predetermined period; The dividing of the first hit number information by the second hit number information to calculate the occupancy rate corresponding to the rising period may include: Calculating a share corresponding to the first rising period Search term for each time characterized in that the method. The method of claim 3, And a part of the search word is excluded from the extracted search word according to a predetermined exclusion criterion. The method of claim 14, The exclusion criterion is related to whether the search term is included in a predetermined name dictionary or whether the search term is included in a predetermined social issue search term dictionary that stores a search term associated with a social issue. . The method of claim 3, The time-specific search word detection system continuously performs the step of claim 3 every predetermined time period, As a result of the execution, the keyword search method for each time period characterized in that the search term detected by the search term by time over a predetermined reference number is determined as the last term search term corresponding to the rising period. The method of claim 16, When a predetermined search term is detected as a time-specific search term corresponding to the first rising period in the first time period and is detected as a time-specific search term corresponding to the second rising period in the second time period, If the difference between the first and second rising periods is within a predetermined error range, determining the detected search term as a time-specific search term corresponding to one third rising period, And the third rising period is determined based on one or more of the first, second rising period, and the difference. A computer-readable recording medium having recorded thereon a program for executing the method of any one of claims 3 to 6, 9, 10, and 12-17. In the system for detecting a time-specific search term that fluctuates the number of inputs from the user at a specific time, A database including input count information per predetermined time interval for the search word for each search word, wherein the database includes the input count information for a predetermined period of time; Input count analyzing means for analyzing the input count information for each keyword by referring to the database; Search term extracting means for extracting a search word in which there is a rising period according to the change of the input count information based on the analysis result, wherein the rising period ends the predetermined rising from the time when the input count information exceeds a predetermined rising start reference value; A time period up to a time point when the temperature falls below a reference value, wherein the rising start reference value and the rising end reference value are determined for each search word; First query information generating means for generating first query information by adding up the number of times of the input information during the rising period; Second inquiry count information generating means for generating second inquiry count information by adding up the number of times of input information for the remaining period except the rising period in the predetermined period; Occupancy calculating means for dividing the first hit number information by the second hit number information to calculate an occupancy rate corresponding to the rising period; And A search word re-extraction means for re-extracting a search word of the extracted search word whose share exceeds a predetermined reference value, The time-based search word detection system, characterized in that for detecting the re-extracted search word as a time-specific search word corresponding to the rising period. The method of claim 19, And a search word re-extraction means for re-extracting a search word having the length of the rising period among the extracted search words existing between a predetermined first reference value and a second reference value. delete

Images