JPH05307569A - Method for storing and retrieving information corresponding to information varied with time - Google Patents

Abstract

PURPOSE:To provide information, which is most suitable for the time period requested by a user, from a data base. CONSTITUTION:This method consists of a relative time calculating part 2 which demarcates a preliminarily designated time range to measure the frequency in occurrence of similar information, a period calculating part 3 which measures and stores the variance of the frequency in occurrence of similar information among plural time ranges, a period calculating part 4 which checks the period of the occurrence of information in accordance with information of the variance of the frequency in occurrence, a data base 1 where information including the period of occurrence of each information is stored, and an information weight setting part 6 which selects information important for the time period from frequencies in occurrence of respective information to the time period given from the user.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of storing and retrieving information corresponding to time-varying information, and more particularly, in a database system for storing and retrieving information.
The present invention relates to a method of storing / retrieving information that separates or selects information that becomes periodically important and sudden information, and a method of accumulating / retrieving information that corresponds to information that changes with time.

[0002]

2. Description of the Related Art In a search by an information search system, particularly when searching for current affairs information, one of the search conditions is, for example, "1" as a search condition in order to limit a temporal area of information to be acquired.
It is possible to give a time range condition such as “January 990.” Even in a system using a paper medium such as a library, by selecting a magazine of a specific year from back numbers of bound magazines, There is a method of performing search by giving a uniform time interval constraint to the entire information.

Thus, conventionally, information generated simultaneously is treated as having the same value with respect to the time axis.

[0004]

However, the value of current affairs information is discussed every year, such as information about the days of the end of the war and typhoon information, or every tenths of a year like Comet Halley. If there is information that has periodic value with respect to time, it is effective at the time when the information occurs, but there is various information such as sudden information that loses value immediately. It is not possible to discard information or limit search in the same time period. However, in the conventional system, in order to treat the information generated at the same time as having the same value with respect to the time axis, it is possible to simply store new information in a high-speed storage medium due to the limitation of the capacity of the storage medium, There is a problem that it is difficult to accurately obtain information that can predict the time when information is needed in advance, such as seasonal topics, simply by removing information that has elapsed.

The present invention has been made in view of the above points,
It is an object of the present invention to provide a method of accumulating and retrieving information corresponding to time-varying information, which can provide optimum information for a time period requested by a user.

[0006]

According to the present invention, the first information that appears periodically, one or more second information that expresses the characteristics of the first information, and the time at which the first information occurs are In the stored database, the number of occurrences of similar first information is divided into pre-designated time ranges and measured, and the variation in the occurrence frequency of similar first information within a plurality of time ranges is measured. The cycle in which the second information occurs is checked from the information on the variation in the occurrence frequency, the third information including the cycle of the occurrence for each second information is stored, and the second information for the time cycle given by the user is stored. The degree of importance is determined for the time period from the frequency of occurrence of, and the first information having high importance is selected.

Further, according to the present invention, in a database in which first information which appears periodically, one or more pieces of second information expressing characteristics of the first information, and a time when the first information occurs are stored, In order to calculate the time period in which the second information occurs, one or more keywords are given to the first information, the occurrence distribution of each keyword with respect to the time period is measured, and the keyword time is calculated according to a calculation formula given in advance. The time period in which the first information having the keyword occurs is determined from the frequency of occurrence for the period.

Further, according to the present invention, in the database in which the first information which appears periodically, one or more second information expressing the characteristics of the first information, and the time when the first information occurs are stored, The distribution of the frequency of occurrence of the keyword with respect to the time period is compared with the distribution of the frequency of occurrence of the keyword given in advance to change the influence on the calculation of the importance of the information having the keyword with respect to the time period.

Further, according to the present invention, the first information, the one or more second information expressing the characteristics of the first information, and the time at which the first information occurs are stored in a database. It is equipped with a database that stores time period information and weight period information together with keywords, and compares preset threshold values and weight data by arranging in order of keyword weights for a given time range and time period. Then, the corresponding time period and weight are ordered in descending order, and the important keywords for the time range and time period are determined.

[0010]

According to the present invention, first, the frequency of occurrence of similar information is measured by dividing it into a predetermined time range such as 1 day, 1 week, 1 year, etc. Examine the occurrence cycle, store each information and information including the information occurrence cycle in the database, and select important information for the time cycle from the occurrence frequency of each information for the time cycle given by the user. ..

Further, in order to calculate the time period of information,
A keyword is given to information, a distribution of occurrence of each keyword with respect to a time period is measured, and a time period in which information having the keyword occurs is determined from a frequency of occurrence of each keyword according to a predetermined calculation formula.

Further, when the frequency of occurrence of the keyword is random with respect to the time period, the influence on similar calculation is reduced, and the feature of the time period in which the keyword occurs is similar to the keyword. Increase the effect on degree calculation.

Furthermore, by using a database that stores keywords, time period information, and weighting information for the time period, particularly one year in a predetermined certain time range and time period, for example, the time range from March to April. It is possible to determine an important keyword in a corresponding time period and time range by arranging the weights of the keywords in order of the period and comparing the data of the threshold and the weight given in advance.

As described above, according to the present invention, by distinguishing information according to the frequency of occurrence of information, information that becomes a periodic topic such as the day of the atomic bomb and transient information such as a sudden incident. It is most important to be able to handle information separately and to identify important keywords corresponding to an arbitrary time range and time cycle, such as a typhoon that is a topic that periodically becomes a topic from August to September. This is a major feature, and in the conventional technology, all information in the search target period was searched equally.
As a result, the information that was in the search target period is accurately searched.

[0015]

Embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows a system configuration of information storage and retrieval according to an embodiment of the present invention.

The system of the present invention comprises a database 1 comprising a current affairs information database 11 and a keyword database 12, a relative time calculation unit 2, a distribution information storage unit 3, a period calculation unit 4, a keyword weight setting unit 5, an information weight setting unit. 6 and a time range determining unit 7.

The database 1 is composed of a current affairs information database 11 for storing current affairs information, a keyword database 12 for storing keywords and relative time of each keyword, and a keyword database 12 for storing absolute time for retrieval.

In the present embodiment, the structure of the current affairs information has data relating to one or more contents, at least one or more keywords, and an information generation time. The time of occurrence of information is
For example, it can be realized by posting each current affairs information in a newspaper or using the date when it is broadcast on TV or radio.

The relative time calculation unit 2 converts the generation time of the information retrieved from the database 1 and the input information from the oldest data stored in the database 1 into a relative time.

The distribution information storage unit 3 has a two-dimensional slot based on the relative time output from the relative time operation unit 2 and each keyword, and the slot corresponding to the slot determined from the time and the keyword determined by the relative time operation unit 2. Increment the value and save.

The period calculation unit 4 determines the period in which information is generated from the distribution of values stored in the distribution information storage unit 3.

The keyword weight setting unit 5 includes a period calculation unit 4
Based on the period information from, the weight for the period of each keyword is calculated.

The information weight setting unit 6 calculates the weight for each current affairs information from the weight information given for each keyword when searching the database 1.

The time range determination unit 7 calculates the width of the time period to be searched.

In FIG. 1, the solid line shows the flow of data registration in the database 1, and the dotted line shows the database 1.
The flow at the time of data retrieval of is shown.

In order to identify the information, each piece of current affairs information entered into the database 1 is provided with a plurality of keywords representing the genre and characteristics of the information and an occurrence time indicating the time when the information occurred.

The distribution information storage unit 3 in this embodiment has, for example, 1 to 31 days, 32 days to 6 as a time range.
Give a time range every month, such as a day. Then, the time given by the relative time calculation unit 2 is divided by 365 days to obtain a remainder, and the leap year is converted to March 1st as the first day to obtain a relative number of days within the range of one year. Convert and increase the value of the corresponding slot. As a result, for example, year-end lottery information is concentrated in slots from 276 days to 306 days, so year-end lottery information is important in December, which is in the range of 276 days to 306 days. It can be judged as information.

FIG. 2 shows a flow chart of a process of registering current affairs information according to an embodiment of the present invention.

Step 20: When new current affairs information is added to the database 1, current affairs information and occurrence time are input.

Step 21: Using each keyword possessed by the new current affairs information, the current affairs information database 11 is searched and the generation time of the information having the corresponding keyword is acquired.

Step 22: Current Affairs Information Database 11
The occurrence time searched for is converted from a predetermined time by the relative time calculation unit 2 into a relative time, and stored in the distribution information storage unit 3 together with the keyword.

The processing from step 23 to step 26 is repeated for the number of keywords.

Step 23: One keyword is retrieved from the distribution information storage unit 3.

Step 24: The period calculation unit 4 obtains the time range of the peak portion from the distribution data for one keyword stored in the distribution information storage unit 3.

Step 25: The keyword weight setting unit 5 compares the distribution value for each keyword with a preset threshold pattern.

Step 26: If the distribution of the keyword with respect to the time period is larger than the threshold value of the peak value, the value is stored in the keyword database 12 together with the value of the period and the value of the phase difference.

As an example of the data stored in the keyword database 12, if the keyword "typhoon" is used, the distribution information storage unit 3 according to the above-described embodiment can be used for "typhoon" and August. The value of the phase difference is 154 to 184 days, the value of the cycle is 365 days corresponding to one year, the value of the weight of August and the value obtained in the same manner for September are stored. This process is repeated as long as there is current information (keyword) to be input.

This process is the same even when there is no data in the database, and becomes stable when sufficient data is stored.

The keyword weight setting unit 5 in this embodiment
As the processing in, the average value is calculated from the slot values of the time range set every January, the weight for the time range less than twice the average value is zero, and the weight for the value more than twice the average value is 1. There is a weight of 3 or more times, and a weight of n or more times the average value is n-1. Further, the value of the weight of each time range is multiplied by the value of the slot for that time range to obtain the weight for that keyword.

As a result, for example, the keyword typhoon appears only once in March or April, but it often appears after June, about 25 times in August and about September. Appears about 40 times. Therefore, on average, they appear about eight times a month, so the weight in August is 25/8 * 25, which is 75, and in September, it is 200.

On the other hand, the keyword stock price appears in about 40 to 60 throughout the year when viewed on a monthly basis, so that the weight becomes 0 in all months and it is not information depending on the time period. Is judged.

Current affairs information database 1 shown in this embodiment
When a user who uses 1 wants to retrieve current affairs information that frequently occurs in a certain season, for example, the user may select the time, date, time range, and time period of the system such as 8 months, 1 month, 1 year cycle, etc. throw into.

FIG. 3 is a flow chart showing a process of a search request from a user who uses the database according to the embodiment of the present invention. Please refer to the flow indicated by the dotted line in FIG.

Step 30: Timing, time range and time period are input.

Step 31: The time range determining unit 7 determines, for example, 1 from the date and time range given in step 30.
From the condition of annual cycle, it is converted into the width of the time cycle to be searched from 350 days to 380 days. This converts the input time cycle into a cycle with the unit of day.

Step 32: The keyword date base 12 is conditioned on the time range determined in step 31 and the time range from 154 days to 184 days, which represents 8 months and 1 month.
To find the relevant keyword and the weight of the keyword. These are important keywords for August in the one-year cycle, and are hereinafter referred to as August keywords.

Step 33: Keyword database 1
One keyword is extracted from the search result of 2, and the following processing is repeated until the processing of all search results is completed.

Step 34: Keyword database 1
The keyword of each information retrieved from 2 is extracted, and the keyword is used to create the current affairs information database 1
Search for 1.

Step 35: In the information weight setting unit 6, for example, the keyword of the current affairs information extracted in Step 34 is compared with the keyword of August obtained in Step 32, and the keyword weight is attached to the matching keyword. Add 0 for keywords that do not match,
The sum of the weight values for each current affairs information for August is calculated.

Step 36: When all the search results have been processed, they are ordered and arranged in descending order of weight created by the information weight setting unit 6 and provided to the user.

FIG. 4 shows the data structure of the keyword database in one embodiment of the present invention. The structure of the keyword database 12 includes at least one or more period information for one keyword. The period information is composed of time period information such as one year obtained by the method of the present invention, phase difference information such as September, and weight information for the time period. In the example of the figure, the keyword “typhoon” has a cycle of 350 to 380 days, has a phase of 154 to 184 days (corresponding to August) from March 1, and has a weight of 75. Information and a cycle of 350 days to 380 days, having a phase of 185 days to 214 days (corresponding to September) as viewed from March 1st, and having a weight of 200.

Next, another embodiment of the present invention will be described.

The distribution information storage unit 3 expresses the slots in the time range as a relative time range in increments of one day from the time of the oldest information stored in the database 1, and the period calculation unit 4 By calculating the autocorrelation function of the distribution obtained in the distribution information storage unit 3 and calculating the time delay, it is possible to calculate the cycle having an arbitrary number of days instead of the fixed time range such as the monthly unit. ..

In the other embodiments described above, the autocorrelation function is used, but it is also possible to use the Fourier transform to obtain an arbitrary period, and by obtaining the power spectrum, the time period in which information is generated. And the frequency of occurrence can be calculated. Furthermore, by using the Fourier transform, it is possible to obtain information on the phase difference as well as information on the period.

Next, another embodiment of the keyword weight setting unit 5 will be described. The logarithmic function log ₂ of the value of the slot in each time range is obtained. When the logarithmic function log ₂ is equal to or more than a threshold value given in advance, the value of the logarithmic function log ₂ is used as a weight.
(However, even if the value of the slot is 0, the weight is 0). As a result, since the weights of the weights are made uniform by the log function in a region having a large weight, the influence of the value of a certain specific keyword does not become too large.

As another embodiment of the keyword weight setting unit 5, when the frequency of occurrence of a keyword is transformed into the frequency domain by using Fourier transform, if the power spectrum is larger than a threshold value given in advance, The value is used as the weight of the keyword, and 0 when the value is smaller than the threshold value.
And Then, the information on the time period is acquired as the relative number of days from the information on the phase difference.

Next, another embodiment for searching the database will be described. The keywords searched in step 32 of FIG. 3 described above are ordered from the one with the larger value using the weight information of the corresponding time range, and the user is provided with information having a weight greater than a preset threshold value. By doing so, it is possible to obtain important keywords in a certain time range.

For example, by searching the database in step 32 in FIG. 3, "typhoon" (weight value 75), "summer" (weight value 100),
"Water shortage" (weight value 50), "Beer" (weight value 3)
0) etc. are obtained, they are arranged in the order of the weight data attached to these keywords. For example, if 50 is given as the threshold value, “summer”, “typhoon”, “water shortage”
It can be obtained by selecting important keywords in August.

The range of the value of the time period given by the user is, in one embodiment, a minimum of one day and a maximum of the current affairs information database 1.
It is the number of days from the date and time of occurrence of the oldest information stored in 1 to the present. The unit for designating the cycle can be not only the number of days but also one week or one year. In this case, it can be achieved by converting the specified cycle into a cycle of standard days and having a width specified in advance. For example, if it is a one-year cycle, it is converted to a standard 365 days, and further, for example, with a width of 5% before and after, it is searched from 347 days to 383 days. Although the unit of the time period in one embodiment is one day, it can be set to an arbitrary value by changing the sampling of the occurrence time of the current affair information.

As described above, the method according to the present invention can identify heavy keywords corresponding to arbitrary time ranges and time periods. With this keyword, more appropriate information can be obtained from the information stored in the database.

[0062]

As described above, according to the present invention, for example,
Temporary information that occurs regardless of the season or time cycle, such as stock price information or traffic accident information, has a smaller weight for the time cycle. On the other hand, information such as festivals becomes less weighted as time passes from the time information is generated, but since related information is repeatedly generated every year, when viewed in the time range of one year cycle, The weight of the information becomes larger than that of the information described above, and it becomes possible to separate the periodically important information from the information that occurs daily or the unexpected information.

As a result, it is possible to provide the optimum information for the time period requested by the user by registering the information with the keyword in the database every day by using the method of the present invention.

[Brief description of drawings]

FIG. 1 is a system configuration diagram of an information storage / retrieval method according to an embodiment of the present invention.

FIG. 2 is a flowchart of a process of registering current affairs information according to an embodiment of the present invention.

FIG. 3 is a flowchart showing a process of processing a search request from a user who uses a database according to an embodiment of the present invention.

FIG. 4 is a diagram showing a data structure of a keyword database according to an embodiment of the present invention.

[Explanation of symbols]

 1 database 2 relative time calculation unit 3 distribution information storage unit 4 cycle calculation unit 5 keyword weight setting unit 6 information weight setting unit 7 time range determining unit 11 current affairs information database 12 keyword database

Claims (4)

[Claims] 1. A database in which first information that appears periodically, one or more pieces of second information that represent characteristics of the first information, and a time when the first information occurs are stored in a database. The number of occurrences of the generated first information is divided into pre-designated time ranges and measured, and the variation in the occurrence frequency of similar first information within a plurality of the time ranges is measured. The cycle in which the second information occurs is checked from the information, the third information including the cycle of occurrence for each second information is stored, and the frequency of occurrence of the second information for the time cycle given by the user is calculated. A method of accumulating and retrieving information corresponding to time-varying information, characterized in that importance is determined for a time period and first information having high importance is selected. 2. A database in which first information that appears periodically, one or more second information that expresses characteristics of the first information, and a time when the first information occurs are stored. In order to calculate the time period in which the second information occurs,
One or more keywords are given to the first information, the distribution of occurrence of each keyword with respect to the time period is measured, and a first formula that has the keyword from the frequency of occurrence of the keyword with respect to the time period is calculated according to a predetermined calculation formula. The method for accumulating and retrieving information corresponding to time-varying information according to claim 1, wherein a time period in which the information is generated is determined. 3. A database in which first information that appears periodically, one or more second information that expresses characteristics of the first information, and a time when the first information occurs are stored. A feature is that the distribution of the frequency of occurrence of the keyword with respect to the time period and the distribution of the frequency of occurrence with respect to the time period given in advance are compared to change the influence on the calculation of the importance of the information having the keyword to the time period A method of storing and retrieving information corresponding to time-varying information according to claim 1. 4. A database in which first information that appears periodically, one or more pieces of second information that express the characteristics of the first information, and a time when the first information occurs are stored. It has a database that stores time period information and weight period information together with keywords, and compares the given threshold value and weight data by arranging in order of keyword weights for a given time range and time period. 2. The information corresponding to the time-varying information according to claim 1, wherein the corresponding time periods and weights are arranged in order from the largest and the important keywords for the time range and time period are determined. retrieval method.