KR101563354B1 - Device for determining of collection ranking, method for determining of collection ranking, and method for providing search service using the same - Google Patents

Abstract

The method of determining a collection ranking according to the present invention comprises the steps of extracting data of a search log in which data on a search action input action of a user and a click action after a search word is stored, Finding a query having periodicity using data of the log, and determining the ranking for the collection using the periodicity of the query.

Description

TECHNICAL FIELD [0001] The present invention relates to a collection ranking determining apparatus, a collection ranking determining method, and a search service providing method using the same. BACKGROUND ART [0002]

The present invention relates to a collection ranking determining apparatus, a collection ranking determining method, and a search service providing method using the same.

Internet users can search for interested companies on a web page of a search service providing system and receive information on the searched companies. At this time, the user's search can be performed by variously inputting queries related to the interested company. That is, the search service providing system stores content that can be provided as a search result, and when a user inputs a query, the search service providing system selects a content for the query and provides it as a search result list.

At this time, the query can be related to at least one collection, and it is difficult to provide search results for all collections related to the query. Accordingly, the user can select a collection rank among a plurality of collections related to a query, and expose corresponding search results.

However, according to this method, the updating of the collection ranking may be slow, or the collection ranking selection may not be precisely selected to suit the user's intention at the present time. Accordingly, a search result that does not match the search intention of the user is provided, which may reduce the satisfaction with the search result of the user.

The present invention proposes a collection ranking determination apparatus, a collection ranking determination method, and a search service providing method using the same, which can more quickly provide search results matching the user's search intention.

The method of determining a collection ranking of the present invention comprises the steps of extracting data of a search log in which data on a user's search word input behavior and a click action after a search word is stored, Finding a query having a periodicity using the data of the query, and determining the ranking for the collection using the periodicity of the query.

The step of finding a query having the periodicity includes the steps of defining a window having at least one time interval, calculating a correlation coefficient from a first starting time point to a time interval of the window and a time interval from the second starting time point to the time interval of the window autocorrelation, and calculating a moving average value of the window from the correlation coefficient.

The step of determining the ranking may include deriving a weight at a specific point in time using the moving average value and deriving a ranking of the collection by reflecting the weight at the specific point in time.

The step of deriving the weighting may comprise forming a pattern graph of the collection for a particular query having periodicity.

The step of deriving the weighting may further include deriving a positive weight as an upward trend if the slope of the pattern graph is positive at a specific time and deriving a negative weight as a downward trend if the slope of the pattern graph is negative can do.

The step of deriving the rankings of the collection may include deriving a rank of the collection using the positive weights or the negative weights.

The collection ranking determining apparatus of the present invention includes a data extracting unit for extracting data of a search log storing data on a user's inputting action of a search word and a click action after inputting a search word, a query for finding a query having a periodicity And a ranking determining unit for determining a ranking of the collection by deriving a pattern for the periodicity of the query and reflecting the derived pattern as a weight.

The query extracting unit may include a correlation coefficient deriving unit for deriving a correlation coefficient from a first starting time point to a time interval of the window and a time interval of the window from the second starting time point, And a periodicity deriving unit for deriving the periodicity of the specific query using at least one of the correlation coefficient or the moving average value.

Wherein the ranking determining unit includes a pattern graph forming unit for forming a pattern graph of at least one or more collections related to the specific query, a weight calculating unit for calculating a weight for the collection at a specific time point through a slope of the pattern graph, And a ranking calculation unit for calculating a ranking of the collection.

The weight calculation unit may derive a positive weight with an upward trend if the slope of the pattern graph is positive at the specific time and derive a negative weight with a downward trend if the slope of the pattern graph is negative.

The ranking calculator may calculate the ranking of the collection through the positive weight or the negative weight.

The search service providing method of the present invention is a method of providing a search service using a ranking of a collection in a search service providing system, the method comprising: searching a query having a periodicity in a search log; Extracting a rank of a collection related to the query through a pattern of a periodicity of a query, extracting a collection related to the specific time using the derived ranking result, and extracting a search result list .

In the search log, data on the search word input behavior of the users and the click behavior after inputting the search word can be stored.

The step of deriving the ranking of the collection may include deriving a correlation coefficient (autocorrelation) of the query having periodicity using the data of the search log, and calculating a moving average value of the window from the correlation coefficient .

The step of deriving the ranking of the collection may further comprise deriving the pattern from the moving average value and determining the ranking of the collection by reflecting the pattern as a weight.

The step of deriving the pattern may include deriving a pattern graph of the collection through the pattern of periodicity. If the slope of the pattern graph is positive at a specific time, a positive weight is derived as an upward trend, And deriving a negative weight with a downward trend if negative, wherein determining the ranking of the collection may comprise determining the ranking of the collection using the positive weight or the negative weight have.

According to the present invention, when providing a search result to a user, a query having a periodicity is searched, ranking of collections related to the query is determined using a weight according to a periodicity at a specific time point, By providing and extracting the results, it is possible to provide an environment in which search results matching the user's search intention can be provided more quickly.

1 is a view showing a collection ranking determining apparatus according to an embodiment of the present invention.
2 is a diagram illustrating a process of finding a query and a collection having periodicity according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating a process of deriving a periodicity for a query or a collection according to an embodiment of the present invention.
4 is a flowchart illustrating a method of determining a collection ranking according to an embodiment of the present invention.
FIG. 5 is a diagram showing a pattern graph for each collection according to an embodiment of the present invention.
6 is a flowchart illustrating a method of providing a search service using a collection ranking determination apparatus according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise. Also, the terms " part, "" module," and " module ", etc. in the specification mean a unit for processing at least one function or operation and may be implemented by hardware or software or a combination of hardware and software have.

A collection ranking determination method, a collection ranking determination method, and a search service providing method using the collection ranking determination method according to an embodiment of the present invention will now be described in detail with reference to FIG. 1 through FIG.

1 is a view showing a collection ranking determining apparatus according to an embodiment of the present invention. At this time, the collection ranking determining apparatus 100 shows only the schematic configuration necessary for the explanation according to the embodiment of the present invention, and is not limited to this configuration.

Referring to FIG. 1, the collection ranking determining apparatus 100 according to an embodiment of the present invention utilizes the periodicity of a query to quickly provide a search result of a collection matching a search intention of a user at a specific point in time And includes a data extracting unit 110, a query extracting unit 120, and a ranking determining unit 130.

Here, the query includes a search term entered by the user in Internet search. A collection is a search division unit classified according to characteristics of search information, and includes a cafe, a blog, a news, an SNS, and a community.

The data extracting unit 110 extracts data from the search log based on the search history of the users for the query. Here, in the search log, data on the search word input behavior of the users and the click behavior after inputting the search word are stored. Accordingly, the data extracting unit 110 can extract data related to a click or a click count of a specific query or a specific collection related to a specific query over time.

The query extracting unit 120 finds a query or a collection having a periodicity using the data of the search log extracted by the data extracting unit 110. At this time, the query extracting unit 120 may extract a periodic click pattern for the collections using the period of the query having the periodicity.

Here, the period or the periodicity indicates that the query is searched with a predetermined time interval, and the click of the document in which the query is input or related thereto is periodically generated at a predetermined time interval.

The query extraction unit 120 includes a correlation coefficient derivation unit 122, a calculation unit 124, and a periodicity derivation unit 126 according to an embodiment of the present invention.

The correlation coefficient derivation unit 122 derives a correlation coefficient (autocorrelation) of the query having the periodicity using the data of the search log. First, the correlation coefficient derivation unit 122 can define windows having a time interval (for example, 10 minutes, 30 minutes, 1 hour, or 3 hours).

The correlation coefficient derivation unit 122 calculates the correlation coefficient from the first start time S to the time interval w of the window and the second start time S min to the time interval w + The correlation coefficient is derived.

Here, the correlation coefficient includes a value obtained by quantifying the relationship between the query and the click generated for each time period or a value obtained by quantifying the relationship between clicks generated by the collection for each time period. Therefore, a large correlation coefficient of a specific query in a specific time interval means that there is a high probability that a specific query will have a certain period of time.

At this time, the correlation coefficient derivation unit 122 repeats the process of deriving the correlation coefficient in the time interval while continuously adding time in units of minutes, and can utilize a value with a maximum value. In addition, the correlation coefficient derivation unit 122 may divide the correlation coefficient in units of seconds to improve performance or reliability.

The calculation unit 124 calculates a moving average value of the window. Here, the moving average value includes the average of the number of clicks or the correlation coefficient obtained by moving over a predetermined time interval according to the change of time. This moving average value can be utilized in deriving the periodicity of the query or determining the ranking of the collection.

Accordingly, the collection ranking determining apparatus 100 according to an embodiment of the present invention can grasp the trend through the graph of the moving average value obtained from the correlation coefficient, obtains the slope at a specific point in the graph of the moving average value, You can use it to determine your collection ranking.

The periodicity deriving unit 126 finds a specific query having periodicity using the values obtained by the correlation coefficient derivation unit 122 or the calculation unit 124 and derives the periodicity of a specific query. That is, the periodicity deriving unit 126 can find a specific query having a period using at least one of a correlation coefficient and a moving average value, and derive the periodicity of the specific query.

The ranking determining unit 130 derives a pattern of a query or a collection having a periodicity, and calculates a weight at a specific time point from the derived pattern to determine a ranking of the collection.

The ranking determining unit 130 includes a pattern graph generating unit 132, a weight calculating unit 134, and a ranking calculating unit 136 according to an embodiment of the present invention.

The pattern graph forming unit 132 forms a pattern graph of the collections related to the query having the periodicity. Here, the pattern graph is a wave graph indicating the degree of clicks of collections with respect to time, which will be described in detail with reference to FIG. 5 below.

The weight calculation unit 134 calculates the respective weights of the collections at a specific time point through the pattern graph formed in the pattern graph formation unit 132. [ At this time, the weight calculation unit 134 may calculate a weight for a specific collection through a gradient at a specific point in time.

Here, the specific point in time includes a point in time when a collection ranking is to be derived, or a point in time when a query having a periodicity of a predetermined time period or a query related to the collection is input. In addition, the weight may indicate a variation trend of clicks generated for each collection at the specific time.

Then, the ranking calculation unit 136 calculates the ranking of the collection by reflecting the weight.

For example, if the slope value of the pattern graph is positive at a specific point in time, the positive weighted value is derived as an upward trend. If the slope value of the pattern graph is negative, the weighted value calculation unit 134 can derive the negative weight as a downward trend have. Then, the ranking calculation unit 136 calculates the ranking of the collection through a positive weight or a negative weight.

Accordingly, when the slope value of the pattern graph of the first collection at the present time is a positive trend, the collection ranking determining apparatus 100 increases the ranking for the first collection by applying a positive weight. Likewise, if the slope value of the pattern graph of the second collection at the current time point is a negative trend downward, the rankings for the second collection are reduced by applying a negative weight.

Now, referring to FIG. 2, a process of determining a collection and a query having a periodicity by the collection ranking determining apparatus 100 will be described in detail.

2 is a diagram illustrating a process of finding a query and a collection having periodicity according to an embodiment of the present invention. At this time, the following flowchart will be described using the same reference numerals in conjunction with the configuration of FIG.

Referring to FIG. 2, the collection ranking determining apparatus 100 extracts data from the search log, and extracts data related to the click or click count from the search log related to the query or the collection (S100).

Then, the collection ranking determining apparatus 100 grasps the periodicity of the query using the data of the search log, and if the query has periodicity with a predetermined time interval, the periodicity of the collections related to the query is grasped. At this time, the collection ranking determining apparatus 100 can grasp the periodicity of the collections using clicks per collection per query (S102, S104, S106).

When the collection has periodicity, the collection ranking determining apparatus 100 periodically measures the analysis values such as the number of clicks according to the time of the query, and applies it to the ranking determination of the collection related to the query (S 108, S110).

Referring now to FIG. 3, a process of deriving a correlation coefficient and deriving a periodicity for a query or a collection will be described in detail.

FIG. 3 is a flowchart illustrating a process of deriving a periodicity for a query or a collection according to an embodiment of the present invention.

Referring to FIG. 3, the collection ranking determining apparatus 100 defines windows having at least one time interval and derives a correlation coefficient for a query or a collection (S200, S202). At this time, the collection ranking determining apparatus 100 repeats the process of deriving the correlation coefficient in the time interval by continuously adding time in units of minutes or seconds, and can utilize a value of the maximum value.

The collection ranking determining apparatus 100 calculates a moving average value of the window from the correlation coefficient for the query or the collection (S204).

Then, the collection ranking determining apparatus 100 derives the periodicity of the query or the collection using at least one of the correlation coefficient or the moving average value (S206). That is, the collection ranking determining apparatus 100 can derive the periodicity of the query or derive the periodicity of the collection through the correlation coefficient. In addition, the collection ranking determining apparatus 100 may derive the periodicity of the query or the periodicity of the collection using the moving average value.

Referring now to FIG. 4, the process of determining the ranking of the collection will be described in detail.

4 is a flowchart illustrating a method of determining a collection ranking according to an embodiment of the present invention.

Referring to FIG. 4, the collection ranking determining apparatus 100 extracts data of a search log and derives a query having a periodicity using data of the extracted search log (S300, S302). The collection ranking determining apparatus 100 extracts click data according to time of a query or a collection from a search log, and derives a query having a periodicity using the extracted click data.

Then, the collection ranking determining apparatus 100 forms a pattern graph of the collection related to the query (S304). At this time, the collection ranking determining apparatus 100 may utilize the correlation coefficient obtained when deriving the periodicity of a query or collection, or may form a pattern graph of the collection using a moving average value or the like.

The collection ranking determining apparatus 100 calculates a weight at a specific time point on the collection through the pattern graph (S306). Here, the collection ranking determining apparatus 100 calculates a weight having an upward trend or a downward trend through a slope of the pattern graph at a specific point in time.

The collection ranking determining apparatus 100 determines the ranking of the collection by reflecting the weight (S308). At this time, the collection ranking determining apparatus 100 may determine rankings among different collections, or may determine ranking of different documents in the same collection. And, it can be variously modified or changed according to the user's request utilizing the search environment or the collection ranking determining apparatus 100. [

The pattern graph for each collection will now be described in detail with reference to FIG.

FIG. 5 is a diagram showing a pattern graph for each collection according to an embodiment of the present invention.

Referring to FIG. 5, the X-axis represents the time and the Y-axis represents the degree of click. And, the collection A is a pattern graph of the news collection for the query Q, and the collection B is the pattern graph of the blog collection for the query Q.

The period of the collection A has the value of the period 1, and the period of the collection B has the value of the period 2. Here, the period 1 and the period 2 are calculated by calculating the correlation coefficient while moving the time interval for each collection.

The collection ranking determining apparatus 100 applies the click trend of the current point in the pattern graph of FIG. 5 and uses it to determine the ranking of the collection. For example, when the pattern graph of the first collection is a downward trend (the slope of the graph is negative) and the pattern graph of the second collection is an upward trend (the slope of the graph is positive) at the present point in time, , A strong weight can be applied to the second collection having an upward trend, so that it can be used for determining the ranking of the collection.

5, at the current point C, although the Y value of the collection A is larger than the Y value of the collection B, the pattern graph of the collection A has a downward trend, and the pattern graph of the collection B has an upward trend, Apply increasing tendency to collection B by applying a stronger weight to collection rankings.

Now, referring to FIG. 6, a process of providing a search service by determining a rank of a collection related to a query having a periodicity will be described in detail.

6 is a flowchart illustrating a method of providing a search service using a collection ranking determination apparatus according to an embodiment of the present invention.

Referring to FIG. 6, a method of providing a search service of a search service providing system according to an embodiment of the present invention includes extracting data from a search log by a collection ranking determining apparatus 100, A collection is derived (S400).

When the query for the Internet search is inputted from the user terminal 300 at a specific time, the search service providing apparatus 200 transmits the query inputted from the user terminal 300 to the collection ranking determining apparatus 100, And requests the ranking of the collections related to the query to the collection ranking determining apparatus 100 (S402, S404).

In addition, the collection ranking determining apparatus 100 derives the collection ranking at a specific point of time using the weights of the collections according to the periodicity of the query, and transmits the collection ranking to the search service providing apparatus 200 (S406, S408).

The search service providing apparatus 200 extracts a search result list using the collection ranking transmitted by the collection ranking determining apparatus 100 and provides the extracted search result list to the user terminal 300 (S410, S412) .

Hereinafter, an example will be described in which the collection ranking determining apparatus 100 determines a query "infinite challenge" and a collection ranking and selects a search result list therefrom.

Here, "Infinite Challenge" is a TV program, but is used as a query for search in a search service providing system. "Infinite Challenge" is aired every Saturday, and the usual type of search is before a TV program airs a trailer to wonder, search for trailers, or last week's broadcast content is searched for the contents of the last week to search for the document.

And, while "Infinite Challenge" is on air, write to the SNS or community to share the present contents, and the user reads and reads the document related to it (for example, what shoes is now reported by.

In addition, after the end of "Infinite Challenge", the media such as news broadcasts the broadcasted contents as soon as possible and the users who use the search service providing system utilize it. Also, at the end of broadcasting, a general user writes a review through a blog, and users who use the search service providing system utilize it.

Therefore, the cycle and order for the query "Infinite Challenge" are as follows. First, a trailer video or blog is searched a lot, and then SNS or community is searched a lot. Then, news is searched a lot, and finally searched in the order of blog or cafe.

Accordingly, the collection ranking determining apparatus 100 and the search service providing apparatus 200 according to an embodiment of the present invention can quickly determine the ranking of a collection by utilizing a periodically appearing period of one week, And expose it to the user quickly.

More specifically, for the user who entered "Infinite Challenge", the collection ranking determining apparatus 100 extracts data of the search log. There are a lot of related queries around the time zone of "Infinite Challenge". And, it has a high correlation coefficient with a cycle of one week.

Accordingly, the collection ranking determining apparatus 100 analyzes the search log and derives a query having a high correlation coefficient. Then, the collection ranking determining apparatus 100 analyzes the search log and analyzes the clicked collection of documents related to the query.

Then, the collection ranking determining apparatus 100 calculates the period of people who click on the news, the blog, and the SNS among the people who input the query, and searches for a time period when the user inputs a search word or broadcasts " Find the time. The collection ranking determining apparatus 100 calculates ranking of collections such as news, blog, and SNS in accordance with the specific time point, and the search service providing apparatus 200 provides the search service promptly using the ranking.

In addition, the search service providing apparatus 200 digitizes the data of the search history related to the query through the Ranking Learning (LTR), and then reflects the data to the user searching at a similar point in time, Can be exposed.

In summary, the query "Infinite Challenge" is a TV program that is being airing every week. When the "Infinite Challenge" is being broadcast, the contents of the current broadcast are shared through the community or the SNS, and at the end of the TV program, Is the fastest in the news. Then, the user who watches the TV program has a repetition cycle in the form of leaving a comment on the blog.

Accordingly, the collection ranking determining apparatus 100 and the search service providing apparatus 200 according to an embodiment of the present invention utilize this feature to search for a proper collection or content, such as at a point in time when the TV starts broadcasting, .

As described above, the collection ranking determining apparatus 100 according to an embodiment of the present invention searches a query having a periodicity when providing a search result to a user, and calculates a collection related to the query using a weight based on the periodicity at a specific point in time To determine the ranking.

The search service providing apparatus 200 extracts and provides search results at a specific point in time by utilizing the search results, thereby providing an environment that can provide search results matching the user's search intention more quickly.

The embodiments of the present invention described above are not implemented only by the apparatus and method, but may be implemented through a program for realizing the function corresponding to the configuration of the embodiment of the present invention or a recording medium on which the program is recorded. Such a recording medium can be executed not only on a server but also on a user terminal.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

110: Data extraction unit 120: Query extraction unit
122: correlation coefficient derivation unit 124: calculation unit
126: periodicity deriving unit 130: ranking determining unit
132: pattern graph forming unit 134: weight calculation unit
136: Ranking calculation unit

Claims (16)

In the manner in which the collection ranking decision device determines the ranking of the collection,
A step of extracting data of a search log storing data on a click action after a user inputs a search word input action and a search word,
Finding a query having a periodicity using the data of the search log, and
Determining a ranking for the collection using the periodicity of the query
/ RTI >
Wherein the step of finding a query having the periodicity comprises:
Defining a window having at least one time interval,
Deriving a correlation coefficient (autocorrelation) from a first starting time point to a time interval of the window and from a second starting time point to a time interval of the window, and
Calculating a moving average value of the window from the correlation coefficient
/ RTI > delete The method of claim 1,
The step of determining the ranking comprises:
Deriving a weight at a specific time point using the moving average value, and
Deriving the ranking of the collection by reflecting the weight at the specific time point
/ RTI > 4. The method of claim 3,
The step of deriving the weight includes:
Forming a pattern graph of the collection for a particular query having periodicity
/ RTI > 5. The method of claim 4,
The step of deriving the weight includes:
If the slope of the pattern graph is positive at a specific time point, deriving a positive weight with an upward trend, and if the slope of the pattern graph is negative, deriving a negative weight with a downward trend. The method of claim 5,
Wherein deriving the ranking of the collection comprises:
And deriving a ranking of the collection using the positive weight or the negative weight. A data extracting unit for extracting data of a search log storing data on a search word input behavior of users and a click behavior after inputting a search word,
A query extracting unit for searching for a query having a periodicity using data of the search log, and
A ranking determining unit which derives a pattern of the periodicity of the query and reflects the derived pattern as a weight to determine a ranking of the collection,
/ RTI >
The query extracting unit extracts,
A correlation coefficient derivation unit for deriving autocorrelation from a first starting time point to a time interval of the window and from a second starting time point to a time interval of the window,
A calculation unit for calculating a moving average value of the window, and
And a periodicity deriving unit for deriving a periodicity of a specific query using at least one of the correlation coefficient or the moving average value,
Wherein the collection ranking determining device comprises: delete 8. The method of claim 7,
The ranking determining unit may determine,
A pattern graph forming unit for forming a pattern graph of at least one or more collections related to the specific query,
A weight calculator for calculating a weight for the collection at a specific time point through a slope of the pattern graph, and
A ranking calculation unit for calculating a ranking of the collection by reflecting the weight,
Wherein the collection ranking determining device comprises: The method of claim 9,
The weight calculation unit may calculate,
Wherein when the slope of the pattern graph is positive at the specific time point, a positive weight is derived as an upward trend, and if the slope of the pattern graph is negative, a negative weight is derived as a downward trend. 11. The method of claim 10,
The ranking calculator calculates,
To the ranking of the collection through the positive weight or the negative weight. In a method in which a search service providing system provides a search service using ranking of a collection,
Finding a query having a periodicity in a search log,
When a query for a search is input at a specific time point, deriving a ranking of a collection related to the query through a pattern for the periodicity of the query,
Extracting a collection related to the specific time point using the derived ranking result, and
And providing a retrieved search result list using the extracted collection,
Wherein deriving the ranking of the collection comprises:
Deriving a correlation coefficient (autocorrelation) of the query having periodicity using the data of the search log, and
Calculating a moving average value of the window from the correlation coefficient;
The search service providing method comprising: The method of claim 12,
In the search log,
A method of providing a search service in which data on a search word input behavior of a user and a click behavior after inputting a search word are stored. delete The method of claim 12,
Wherein deriving the ranking of the collection comprises:
Deriving the pattern from the moving average value, and
And determining the ranking of the collection by reflecting the pattern as a weight. 16. The method of claim 15,
Wherein deriving the pattern comprises:
If a slope of the pattern graph is positive at a specific time point, a positive weight is derived as an upward trend, and if the slope of the pattern graph is a negative trend, a negative weight , ≪ / RTI >
Wherein determining the ranking of the collection comprises:
And determining the ranking of the collection using the positive weight or the negative weight.

Images