JP4821135B2 - Data analyzer - Google Patents

Description

  The present invention relates to a data analysis apparatus using a neural network.

  As a data analysis method using a neural network, several methods have been conventionally known. Among them, the one proposed by Hopfield (JJHopfield) is that the coupling strength between the nodes is defined for a plurality of nodes, and the pattern including this node is a predetermined pattern group. It is to determine which is closest.

Here, the coupling strength between the nodes is set to be a so-called symmetric coupling in which the coupling strength from the node i to the node j is equal to the coupling strength from the node j to the node i. It is known that the result of activity propagation in a network of such nodes (symmetric coupling network) settles to a specific stable state (a fixed point that is an attractor of a dynamic system defined by activity propagation). In other words, Hopfield's proposed method determines which network's initial pattern is closest to which fixed point pattern (predetermined pattern) depending on which stable state (fixed point) the network eventually settles in. The calculation is performed (Non-Patent Document 1).
Hopfield, JJ (1982) Proc. Natl. Acad. Sci. USA, 79, 2554 Koulakov, AA, Raghavachari, S., Kepecs, A. & Lisman, JE (2002), Nat. Neurosci. 8, 775 Goldman MS, Levine JH, Major G., Tank DW, Seung HS, "Robust persistent neural activity in a model integrator with multiplehesteretic dendrites per nuron", Cereb. Cortex 13, 1185-1195 (2003)

  However, in the conventional hop field method, the network analysis result is one of predetermined fixed point patterns. These fixed point patterns exist discretely in the solution space, and there is a problem that a solution that continuously changes with respect to the change of the initial pattern cannot be obtained.

  Therefore, an attempt has been made to realize a dynamic system having a continuous attractor by configuring a network with hysteresis units (units having hysteresis characteristics in change in value) (Non-patent Document 2). However, in this Kurarov method, the continuity of the pattern is not realized in the end, and the continuity regarding the scalar quantity (called activity) defined from the whole network is only realized.

  That is, no method has been found for providing a solution having continuity with respect to the initial pattern state of a network configured using a group of data elements to be analyzed. For this reason, any of the above conventional methods can only perform data analysis for limited applications.

  The present invention has been made in view of the above circumstances, and provides a solution having continuity with respect to the initial pattern state of a network configured using a data element group to be analyzed, and is used for a wide range of applications. It is an object of the present invention to provide a data analysis apparatus that can perform the above.

  Another object of the present invention is to provide a data analysis apparatus that reduces the processing load.

  The present invention for solving the problems of the above-described conventional example is a data analysis apparatus, and is a continuous value between Ymin and Ymax (Ymin <Ymax) for at least some of the plurality of data elements. A storage means for storing data values in association with each other, storage means for storing relationship weighting information between the data elements, and at least one of the data elements as a seed, the seed and other data elements different from the seed Means for selecting at least one as a calculation target, selecting one of the selected calculation targets as attention data based on a predetermined rule, the attention data, and each data element selected as the calculation target; The input stimulus value related to the attention data is calculated based on the relationship weighting information and the data value of the data element selected as the calculation target. And means for updating the data value of the data of interest based on the calculated input stimulus value, the threshold value being determined according to the current data value related to the data of interest, and different for each data value A data value updating means for comparing the value and the input stimulus value to determine whether or not to change the data value, and updating the data value when it is determined to change the data value, The data value for at least one of the data elements after repeated execution of the calculation of the input stimulus value and the update of the data value until a condition is satisfied is subjected to a predetermined process. It is characterized by.

  Here, it is possible to further include means for additionally selecting a data element to be calculated based on the data value updated by the data value updating means.

  Further, the means for updating the data value is a first monotonically increasing function in which the input stimulus value is Ymin from 0 to X1min, continuously monotonically increases from X1min to X1max, and becomes Ymax above X1max. Using f1 and the second monotonically increasing function f2 in which the input stimulus value continuously increases monotonically between X2min and X2max and becomes Ymax above X2max, the input stimulus value for the data of interest ranges from X1min to X2max. In the meantime, with respect to the data value Y of the attention data before the update, the data value is updated to f2 (I) when the input stimulus value I exceeds I2 with reference to I2 where Y = f2 (I2) Referring to the data value Y of the attention data before update, referring to I1 where Y = f1 (I1), and updating the data value to f1 (I) when the input stimulus value I is less than I1 It is good.

  Further, the means for updating the data value may perform scaling of the data value so that when the data value is updated, the sum of the data values for each data element becomes a predetermined value.

  The data processing method according to one aspect of the present invention relates to a data value that is a continuous value between Ymin and Ymax (Ymin <Ymax) for each of a plurality of data elements, and stores the data elements in association with each other. Using a computer having storage means for storing relationship weighting information between the data, selecting at least one of the data elements as a seed and selecting at least one of the seed and another data element that is different from the seed as an operation target And selecting one of the selected computation targets as attention data based on a predetermined rule, relation weighting information between the attention data and each data element selected as the computation target, and computation A step of calculating an input stimulus value related to the data of interest based on each data value of the data element selected as a target; A means for updating the data value of the data of interest based on the inputted input stimulus value, the threshold value being determined according to a current data value related to the data of interest and different for each data value; Comparing the stimulus value to determine whether or not to change the data value, and if it is determined to change, updating the data value until the predetermined condition is satisfied It is characterized in that the calculation of the stimulus value and the update of the data value are repeatedly executed, and the data value after the execution is output.

  The program according to another aspect of the present invention stores a data value that is a continuous value between Ymin and Ymax (Ymin <Ymax) in association with each of a plurality of data elements, and stores the data values between the data elements. A step of selecting at least one of the data elements as a seed and selecting at least one of the seed and another data element different from the seed as a calculation target in a computer comprising storage means for storing the relation weighting information And selecting one of the selected calculation targets as attention data based on a predetermined rule, relation weighting information between the attention data and each data element selected as the calculation target, and the calculation target A procedure for calculating an input stimulus value related to the data of interest based on each data value of the data element selected as A means for updating the data value of the data of interest based on the input stimulus value, the threshold value being determined according to the current data value associated with the data of interest and different for each data value, and the input stimulus value To determine whether or not to change the data value, and when it is determined to change, the procedure for updating the data value until the predetermined condition is satisfied. It is characterized in that the above calculation and the update of the data value are repeatedly executed.

  Furthermore, the data analysis apparatus according to an aspect of the present invention includes a storage unit that stores at least a part of a plurality of data elements in association with data values, and stores relation weighting information between the data elements. , One of the data elements is selected as attention data based on a predetermined rule, relation weighting information between the attention data and each data element selected as the calculation target, and selected as the calculation target Means for calculating an input stimulus value related to the data of interest based on each data value of the data element, and means for updating the data value of the data of interest based on the calculated input stimulus value, A threshold value that is determined according to the current data value related to the data of interest and is different for each data value is compared with the input stimulus value to change the data value. Data value updating means for updating the data value when it is determined to change the data value, and calculating the input stimulus value until the predetermined condition is satisfied, and the data value After repetitively executing the update, the relative distance between each data element is determined based on the data value of at least one of the data elements, and the graphic element corresponding to each data element is determined. It is characterized by being arranged and displayed so as to be separated by a relative distance.

  Embodiments of the present invention will be described with reference to the drawings. As shown in FIG. 1, the data analysis apparatus according to the embodiment of the present invention includes a control unit 11, a storage unit 12, a data input unit 13, and a result output unit 14.

  The control unit 11 can be realized by a CPU or the like, and executes data analysis processing according to a program stored in the storage unit 12. The specific contents of this process by the control unit 11 will be described in detail later. The storage unit 12 can be realized using a memory element or a disk device. The storage unit 12 stores a program executed by the control unit 11. The storage unit 12 also operates as a work memory that holds various data used in the data analysis processing of the control unit 11.

  The data input unit 13 and the result output unit 14 are input / output interfaces that acquire data from, for example, a data storage that stores data to be analyzed, and output the analysis results to the data storage or display.

  Next, data analysis processing by the control unit 11 of the present embodiment will be described. In the present embodiment, an interconnection network type data structure is defined, and a data value storage unit having multi-stage hysteresis input / output characteristics is associated with each node. In the present embodiment, the data to be analyzed includes a plurality of data elements. As a specific example, if the data to be analyzed is a collection of papers, data elements can be defined as individual papers.

  The multistage hysteresis input / output characteristics of each node are as follows. First, as shown in FIG. 2A, in the simple hysteresis characteristic (single hysteresis characteristic), the data value Y is either Y = 0 or Ymax. A lower threshold value Qmin and an upper threshold value Qmax are determined. Assuming that the data value Y is initially initialized to “0”, the state of “0” is maintained until the input I exceeds the upper threshold value Qmax. When the input I exceeds the upper threshold value Qmax, the data value Y becomes Ymax. After that, while the input I exceeds the lower threshold value Qmin, the value of Ymax is maintained. When the input I falls below the lower threshold value Qmin, the data value Y becomes “0”.

  Multi-stage hysteresis is obtained by accumulating this single hysteresis characteristic in multiple stages. Specifically, in this multi-stage hysteresis characteristic, each node takes one of N-stage values from Y = 0 to Y = Ymax. Specifically, as shown in FIG. 2B, a lower threshold value Qimin and an upper threshold value Qimax are defined for each stage.

  When the data value is reset to “0” and the input exceeds the upper threshold value Q1max and does not exceed the upper threshold value Q2max in the second stage (when Q1max <I <Q2max), The output is the first stage output Y = Ymax / N. If the input I changes from here to below the lower threshold value Q1min of the first stage, the data value Y becomes “0”.

  When the data value is Y = Ymax / N and the value of the input I exceeds the upper threshold value Q2max in the second stage and does not exceed the upper threshold value Q3max in the third stage (Q2max <I <In the case of Q3max), the data value Y is the second stage value, Y = 2 × Ymax / N.

  As shown in FIG. 3, the control unit 11 generates an element database in which information (for example, a title of a paper) for identifying each data element and a data value are associated with each other, and stores the element database in the storage unit 12.

  The storage unit 12 stores information that defines a network between data elements. As shown in FIG. 4, this information associates information for specifying a data element (for example, a title of a paper) and information for specifying another data element having a connection relationship with the information. Here, the connection relationship may be directional. For example, when a data element is a paper, there are papers that “cited” other data elements and papers “cited” by other data elements, depending on the citation relationship. Therefore, the connection relationship is divided into “citation” and “cited”, and the list of papers cited by itself and the list of papers cited by itself are stored in association with the paper titles and the like.

  Further, as shown in FIG. 5, the control unit 11 generates a relation weighting database in which a relation weight value is set for each pair of data elements, and stores the relation weighting database in the storage unit 12. In the example of FIG. 5, the lower triangular portion is omitted because the lower triangular portion is set to a value obtained by folding the upper triangular portion with a diagonal line (that is, Tij = Tji). In the present embodiment, Tij may be G (a value of G> 0) when there is a citation relationship, and “0” when there is no citation relationship, but is not limited thereto.

  Further, in the present embodiment, the storage unit 12 stores the lower threshold value and the upper threshold value of each step for defining the multistep hysteresis characteristics.

  The control unit 11 performs a data analysis process based on the information stored in the storage unit 12. In this embodiment, such a hysteresis unit is made to correspond to each node in the interconnection network, and active propagation is performed by a “dynamic system” determined in the following steps. In the network comprising the data value storage units of the present embodiment, the attractor (fixed point state) obtained by the active propagation continuously depends on the initial state.

  Hereinafter, the process of activity propagation in the control unit 11 will be described with reference to FIG. First, the control unit 11 performs an initialization process such as selecting an operation target (S1). The control unit 11 causes the user to select one data element that is initially data of interest. Then, for the attention data selected by the user (hereinafter, the initially selected attention data is referred to as initial seed data), a list of data elements in a citation relationship (quotation / cited relationship) is read. Data elements listed in the list of data elements read out here and the initial seed data (that is, a group of data elements adjacent to the initial seed data on the network defined between the data elements), calculation Treated as a target.

  The control unit 11 also initializes the data value for each data element and initializes the element database. As a specific example of the initialization, a predetermined initial value may be set for the initial seed data, and “0” may be set for data elements other than the initial seed data.

  Next, the control part 11 selects one data element as attention data based on a predetermined rule among several (M pieces) data elements used as calculation object (S2). This rule (selection rule) is performed at random, for example. Here, it is assumed that the i-th data element is selected as the data of interest.

  The control unit 11 calculates the input stimulus value of the attention data by the following equation (1) based on the relation weighting information with the data element that is the calculation target (S3).

  Here, Tij is a value of the relation weight between the i-th data element and the j-th data element, and is read from the relation weight database and used. Yj is a data value stored in the element database in association with the jth data element.

  Next, the data value associated with the i-th data element that is the data of interest is determined (S4), and the element database is updated. Specifically, the determination of the data value of the present embodiment is performed as follows.

  That is, it is checked which level n of the multilevel values the data value associated with the data of interest is. Here, since Ymax / N = ΔY is changed every step, the value n of the step can be obtained by dividing the data value by ΔY (that is, since Y = n × ΔY, this is expressed as ΔY. To obtain n). Next, the lower threshold value θmin and the upper threshold value θmax corresponding to the obtained stage are acquired.

  Then, the input stimulus value I calculated in the process S3 is compared with the upper threshold value θmax. If I> θmax, the data value Y is updated to (n + 1) × ΔY, and the process S3 is recursively performed. Run it. In the uppermost stage Y = Ymax, the upper threshold value is set larger than the maximum value of the input stimulus value so that the data value does not exceed the maximum value. Further, the lower threshold value θmin is compared with the input stimulus value I. When I ≦ θmin, the data value Y is updated to (n−1) × ΔY, and the process S3 is recursively executed. Also in this case, in the lowest stage Y = Ymin, the lower threshold value is set smaller than the minimum value of the input stimulus value so that the data value does not fall below the minimum value. If θmin <I ≦ θmax, the process proceeds to step S5.

を演算し、データ値Ｙiに代えて、このデータ値Ｙ′iを記憶部１２に上書きして格納する。このスケーリングによって、ネットワークの収束性が向上される。 The control unit 11 scales the data values so that the sum of the data values for each data element to be calculated becomes a predetermined value C (S5). That is, if the i-th data value is Yi,

And the data value Y′i is overwritten and stored in the storage unit 12 instead of the data value Yi. This scaling improves network convergence.

を生成する（Ｓ６）。そして記憶部１２に前回生成したベクトルが格納されているときには、当該前回生成したベクトルと当該ベクトルとの差を算出し、当該算出された差のベクトルがゼロベクトル（その要素がすべて「０」となっているベクトル）か否か、つまり、

となっているか否かを調べる（Ｓ７）。なお、前回生成したベクトルと今回生成したベクトルとの次元が異なる場合は、処理Ｓ８へ移行する。また、次元が同じ場合であっても、この処理Ｓ７での差ベクトルが、ゼロベクトルでない場合（Ｎｏの場合）は、処理Ｓ８に移行して、演算対象の拡大処理を行う（Ｓ８）。 After updating and scaling the data values associated with the data of interest in this way, a vector defined by the set of data values for each data element

Is generated (S6). When the previously generated vector is stored in the storage unit 12, the difference between the previously generated vector and the vector is calculated, and the calculated difference vector is a zero vector (all elements are “0”). Or not), that is,

It is checked whether or not (S7). If the dimension of the vector generated last time and the vector generated this time are different, the process proceeds to step S8. Further, even if the dimensions are the same, if the difference vector in this process S7 is not a zero vector (in the case of No), the process proceeds to process S8, and the calculation target enlargement process is performed (S8).

  In this process S8, the control unit 11 additionally selects a data element to be calculated based on the updated data value. A specific processing example will be described below. The control unit 11 refers to the data value of the data element other than the initial seed data among the data elements to be calculated, and the data element exceeds a predetermined weight threshold (for example, “0”). Read a list of data elements that are adjacent to the data element (listed as having a connection relation), concatenate to the current list of data elements to be calculated, and increase the number of data elements to be calculated . Here, the connection means that the connection is performed excluding the overlap, which is equivalent to the process of calculating the sum of the sets.

  Note that, when performing the connection, the control unit 11 may control not to perform the connection when the number of data elements to be calculated exceeds a predetermined maximum value after the connection. Thereby, it is possible to prevent the number of data elements to be calculated from increasing without limit.

  After performing the enlargement process of the calculation target, the control unit 11 returns to process S2 and repeats the process. Here, after the data value is updated, it is checked whether or not the data value related to each data element is changed, and the process returns to step S2 while the data value is changed.

  If the calculated difference vector is a zero vector in the process S7 (Yes), the process ends. At this time, a set of information and data values specifying the data elements stored in the element database is provided as a result of analysis to a later predetermined process. The state stored in the element database represents a state in which the input / output of each data element is balanced (so-called dynamic system attractor). This state corresponds to a state called “fixed point”.

  Further, in the calculation in the process S4 described here, it is necessary to execute the calculation process recursively. That is, the data value may change by a plurality of stages. In this way, when changing over a plurality of stages, the control unit 11 repeatedly performs a comparison operation between the threshold value and the input stimulus value. If the number N of stages is increased, the calculation speed is reduced. Therefore, in this embodiment, instead of discretely setting the data value to N stages, the data value is handled as a continuous value assuming the limit when N is infinite. The continuous value here is expressed as a discrete value as the internal representation of the computer, but in theory it is called a continuous value in the sense that it does not take a stepwise value. It is.

In the case of a continuous value, as shown in FIG. 7, the input stimulus value is Ymin from 0 to X1min, continuously increases monotonically from X1min to X1max, and becomes Ymax when X1max or more. Data value update processing is performed using the monotonically increasing function f1 and the second monotonically increasing function f2 in which the input stimulus value continuously increases monotonically between X2min and X2max and becomes Ymax above X2max. That is,
(1) When the input stimulus value I for the data of interest exceeds X2max, that is, when X2max <I:
The data value is Y = Ymax.
(2) When the input stimulus value I for the data of interest is X1 min or less, that is, when X1min <I:
The data value is Y = Ymin.
(3) When the input stimulus value I for the data of interest is between X1min and X2max, that is, when X1min <I ≦ X2max:
Regarding the data value Y of the attention data before update, referring to I2 where Y = f2 (I2), if the input stimulus value I exceeds the I2, the data value is updated to f2 (I),
With respect to the data value Y of the attention data before update, with reference to I1 where Y = f1 (I1), when the input stimulus value I is less than I1, the data value is updated to f1 (I).

として定義できる。この場合、Ｙ＝ｆ２（Ｉ２）、及びＹ＝ｆ２（Ｉ１）を、それぞれ、Ｉ２，Ｉ１について解いた解が

となる。従って、上記データ値の更新処理は、次のようになる。すなわち、
（１）注目データに対する入力刺激値ＩがＸ２maxを超える場合、つまりＸ２max＜Ｉの場合：
データ値Ｙ＝Ｙmaxとする。
（２）注目データに対する入力刺激値ＩがＸ１min以下の場合、つまりＸ１min＜Ｉの場合：
データ値Ｙ＝Ｙminとする。
（３）注目データに対する入力刺激値ＩがＸ1minからＸ2maxまでの間の場合、つまりＸ１min＜Ｉ≦Ｘ２maxの場合：
更新前の注目データのデータ値Ｙについて、入力刺激値ＩがＹ／α＋Ｘ２minを超える場合に、データ値をｆ２（Ｉ）＝α（Ｉ−Ｘ２min）に更新する。また更新前の注目データのデータ値Ｙについて、入力刺激値ＩがＹ／α＋Ｘ１min未満となる場合に、データ値をｆ１（Ｉ）＝α（Ｉ−Ｘ１min）に更新する。この方法によると、判断処理の回数を制限でき、計算量を低減できる。 Here, the first and second monotonically increasing functions f1 and f2 are

Can be defined as In this case, Y = f2 (I2) and Y = f2 (I1) are solved for I2 and I1, respectively.

It becomes. Therefore, the data value update process is as follows. That is,
(1) When the input stimulus value I for the data of interest exceeds X2max, that is, when X2max <I:
The data value is Y = Ymax.
(2) When the input stimulus value I for the data of interest is X1 min or less, that is, when X1min <I:
The data value is Y = Ymin.
(3) When the input stimulus value I for the data of interest is between X1min and X2max, that is, when X1min <I ≦ X2max:
When the input stimulus value I exceeds Y / α + X2min for the data value Y of the attention data before update, the data value is updated to f2 (I) = α (I−X2min). When the input stimulus value I is less than Y / α + X1min for the data value Y of the attention data before update, the data value is updated to f1 (I) = α (I−X1min). According to this method, the number of determination processes can be limited, and the amount of calculation can be reduced.

として、入力刺激値に引用・被引用数に係る係数を乗じてもよい。（２）式において、ki(in)は、ｉ番目のデータ要素である論文が引用している、他のデータ要素の総数、ki(out)は、ｉ番目のデータ要素である論文を、引用している他のデータ要素である論文の総数を意味している。つまり、ここでは引用関係の総数のうち、自己が引用している分がどれだけであるかを係数としていることになる。この係数は、ネットワークを規定する情報を参照して定めることができる。 In the above description, when the data element is, for example, a paper, many other papers are cited. However, papers that are not cited from others result in relatively high activity. Therefore, in order to avoid this, instead of equation (1),

As an alternative, the input stimulus value may be multiplied by a coefficient relating to the number of citations and citations. In equation (2), ki (in) is the total number of other data elements cited by the paper that is the i-th data element, and ki (out) is the paper that is the i-th data element. It means the total number of papers that are other data elements. In other words, here, out of the total number of citation relationships, how much is quoted by the self is used as a coefficient. This coefficient can be determined with reference to information defining the network.

  Next, the operation of the data analysis apparatus according to this embodiment will be described. Here, there is a paper that becomes a seed document, and the process of searching another paper with similar content to this paper is described as an example.

  The control unit 11 extracts words that satisfy a predetermined condition from words extracted from the paper that has become the seed document. Here, the predetermined conditions include, for example, those that frequently appear in the paper (so-called TFIDF method). Then, the control unit 11 searches the paper collection for at least one paper that is close to the seed document using the word extracted from the seed document. This search is performed by a method using a widely known keyword search or word appearance frequency. It is also possible to display the articles found by the search and allow the user to select at least one of them.

  The control unit 11 performs the process shown in FIG. 6 using the paper found by the search here or the paper selected by the user as the initial seed data.

  That is, a data group given to the initial seed data is actively propagated with a paper group having a citation relationship with the initial seed data as an initial calculation target. Then, after active propagation, refer to the data value of each data element included in the operation target, and for the data element whose data value exceeds the predetermined weight threshold, specify the paper group that has a citation relation to it In addition to the calculation target, the calculation target is expanded. Then, the activity propagation process is continued.

  When the process of activity propagation converges, the control unit 11 lists the papers that are each data element in descending order of data values and presents them to the user.

  In this embodiment, in order to use such activity propagation, a solution having continuity with respect to the state of the initial pattern is given to the network configured using the data element group to be analyzed. For this reason, it can be used for a wider range of applications than a network that converges into a solution that can be determined in a discontinuous manner.

  Furthermore, in the present embodiment, in the activity propagation process, data value scaling is performed (normalization is performed by dividing each data value by the sum value and multiplying it by C to obtain a constant sum C). Therefore, the convergence is improved without damaging the network structure. In this case, if the convergence is higher when the scaling is not performed based on the initial value of the data value, the user may be able to appropriately select whether or not to perform the scaling. When the user selects not to perform the scaling, the control unit 11 proceeds to the process S6 after the process S4 without executing the process S5 in the process of FIG.

  Furthermore, in the present embodiment, since the calculation target is limited by a predetermined network structure and the calculation target is expanded by active propagation, the scale of the calculation can be limited and the processing load can be reduced.

  Furthermore, although only the direction in which the operation target is expanded is shown here, for example, in the process of active propagation, for a data element whose data value falls below a predetermined lower threshold, the data element may be removed from the operation target. Good. In this case, for data elements that lose link with the remaining network after the data element is removed (that is, other data elements that are connected to the network only through the removed data element) In addition, it may be removed from the object of calculation.

  In this embodiment, when the data element is a paper, the network structure is defined by the citation relationship. However, the identity of the author (whether or not the same author is included), the appearance frequency of the word, the paper The network structure may be defined from the access log to the collection (whether there is a possibility of being accessed at the same time) or the like. Also, the value of the relation weight may be determined based on information such as the identity of the author, the appearance frequency of the word, and the access log.

  Further, the network may be configured as follows.

(1) Network between Web pages In the case of a document in which links are set mutually, such as a web page, the relationship of the links may be directly used as a network structure.

(2) Relationship between authors In addition, a network may be defined by setting a link between copyrighted works shared by authors. For example, when the author X is the author of the documents A and B and the author Y is the author of the documents A and C (in this case, the authors X and Y are co-authored with respect to the document A), A network having a link between documents A and C is formed.

(3) Co-authoring relationship Further, in the case described above, a network may be defined by setting a link between authors who have a joint work with the author as a data element. In the above case, since authors X and Y have a joint work of document A, a network having a link between authors X and Y is formed.

(4) Relationship between author and document Data elements included in the network are not necessarily the same type. For example, the author's data element and the document data element may be linked in a work relationship. For example, if the author X is the author of the documents A and B and the author Y is the author of the documents A and C as in the above example, the data element of the author X and the data element of the document A (hereinafter, Links are set between author X and document A, between author X and document B, between author Y and document A, and between author Y and document C. . In this case, there is no direct link between the documents A and B, but the relationship is connected via the author X.

(5) Access history For each user, a network may be defined by setting a link between documents used by the user. For example, if there is a history that user X has accessed document A and document B, a link is set between the data element of document A and the data element of document B.

(6) Access to a common document A network may be defined by setting a link between users who have accessed a common document. For example, when the user X is accessing the documents A and B and the user Y is accessing the documents A and C, there is a record of accessing the same document A. To form a network.

(7) For each user, a network may be defined by setting a link between the user and a document accessed by the user. For example, if there is a history that user X has accessed document A and document B, between user X's data element and document A's data element, and between user X's data element and document B's data element, Set a link for each.

(8) Application Document Network Furthermore, when the inventor X is the inventor of the patent applications A and B, a link may be set between the patent application documents A and B to form a network of patent application documents.

(9) Joint Inventor Network Similarly, a link may be set between joint inventors to form an inventor network. That is, when the inventor X and the inventor Y jointly invent the invention A, a link is set between the data elements representing each of the inventors X and Y.

(10) Inventor-Application Network Further, a link may be set between the inventor and the patent application. For example, when the inventor X is the inventor of the patent applications A and B, a link is set between the data element of the inventor X and the patent application documents A and B, respectively. In this case, for example, when the joint inventor Y exists for the invention related to the patent application document A, a link is not set directly between the inventor X and the inventor Y, but the data element of the patent application document A is The link will be set via

  As described above, these network presentation methods include, in addition to the method of presenting data values in descending order, and after the active propagation has converged, the relative distance between each data element is determined based on each data value. Graphic elements (for example, rectangles, circles, etc.) representing data elements may be arranged and displayed so as to be separated by a distance corresponding to the above-described relative distance. In addition, data elements to which links are set may be displayed by being connected with line segments. Further, for example, when different data elements such as “inventor” and “patent application document” are included, various data elements may be displayed using different graphic elements. Specifically, this display is as shown in FIG. According to this display, for example, patent application documents having large data values or inventors are arranged close to each other. From now on, for example, it is possible to analyze information such as a human relation between a certain technology and a certain technology. Further, the color density of each graphic element may be changed according to the data value.

1 is a configuration block diagram of a data analysis apparatus according to an embodiment of the present invention. It is explanatory drawing showing the example of the update method of the data value for every data element. It is explanatory drawing showing the example of the content of an element database. It is explanatory drawing showing the example of the information which prescribes | regulates the network between data elements. It is explanatory drawing showing the example of the content of a relation weight database. It is a flowchart figure showing the processing example of activity propagation. It is explanatory drawing showing the example of the update method of a data value. It is explanatory drawing showing the example of the display method of a data value.

Explanation of symbols

11 control unit, 12 storage unit, 13 data input unit, 14 result output unit.

Claims (5)

For at least some of the plurality of data elements, a data value that is a continuous value from Ymin to Ymax (Ymin <Ymax) is stored in association with each other, and relational weight information between the data elements is stored. Means,
Means for selecting at least one of the data elements as a seed, the seed, and at least one of other data elements different from the seed as an operation target;
One of the selected computation objects is selected as attention data based on a predetermined rule, relation weighting information between the attention data and each data element selected as the computation object, and selection as a computation object Means for calculating an input stimulus value related to the data of interest based on each data value of the generated data element;
Based on the calculated input stimulus value, a means for updating the data value of the data of interest, which is determined according to the current data value related to the data of interest, and a different threshold for each data value; A data value update unit that compares the input stimulus value to determine whether to change the data value, and updates the data value when it is determined to change;
Including
The means for updating the data value is:
When the input stimulus value is 0 to X1min, it is Ymin, continuously increases monotonically from X1min to X1max, and when X1max or more, the first monotonically increasing function f1 becomes Ymax, and the input stimulus value is from X2min to X2max And a second monotonically increasing function f2 that continuously increases monotonically during the period, and becomes Ymax above X2max,
When the input stimulus value for the data of interest is between X1min and X2max,
Regarding the data value Y of the attention data before update, referring to I2 where Y = f2 (I2), if the input stimulus value I exceeds the I2, the data value is updated to f2 (I),
Regarding the data value Y of the attention data before update, referring to I1 where Y = f1 (I1), when the input stimulus value I is less than this I1, the data value is updated to f1 (I),
The data value for at least one of the data elements after repeatedly performing the calculation of the input stimulus value and the update of the data value until the values of the plurality of data elements no longer change is a predetermined value. A data analyzer characterized by being used for processing. The data analysis device according to claim 1,
A data analysis apparatus further comprising means for additionally selecting a data element to be calculated based on the data value updated by the data value update means. The data analysis device according to claim 1 or 2 ,
The data analysis apparatus characterized in that the means for updating the data value scales the data value so that the sum of the data values for each data element becomes a predetermined value when the data value is updated. Computer
For at least some of the plurality of data elements, a data value that is a continuous value from Ymin to Ymax (Ymin <Ymax) is stored in association with each other, and relational weight information between the data elements is stored. and means,
Means for selecting at least one of the data elements as a seed, the seed, and at least one of other data elements different from the seed as an operation target;
One of the selected computation objects is selected as attention data based on a predetermined rule, relation weighting information between the attention data and each data element selected as the computation object, and selection as a computation object Means for calculating an input stimulus value related to the data of interest based on each data value of the generated data element;
Based on the calculated input stimulus value, a means for updating the data value of the data of interest, which is determined according to the current data value related to the data of interest, and a different threshold for each data value; A data value update unit that compares the input stimulus value to determine whether to change the data value, and updates the data value when it is determined to change;
Function as
When functioning as a means for updating the data value,
When the input stimulus value is 0 to X1min, it is Ymin, continuously increases monotonically from X1min to X1max, and when X1max or more, the first monotonically increasing function f1 becomes Ymax, and the input stimulus value is from X2min to X2max And a second monotonically increasing function f2 that continuously increases monotonically during the period, and becomes Ymax above X2max,
When the input stimulus value for the data of interest is between X1min and X2max,
With respect to the data value Y of the attention data before update, referring to I2 where Y = f2 (I2), when the input stimulus value I exceeds I2, the data value is updated to f2 (I),
For the data value Y of the data of interest before update, referring to I1 where Y = f1 (I1), if the input stimulus value I is less than I1, the data value is updated to f1 (I),
The program for repeatedly executing the calculation of the input stimulus value and the update of the data value until the values of the plurality of data elements no longer change . Storing at least a part of the plurality of data elements in association with data values, respectively, and storing relationship weighting information between the data elements;
One of the data elements is selected as attention data based on a predetermined rule, relation weighting information between the attention data and each data element selected as the calculation target, and data selected as the calculation target Means for calculating an input stimulus value related to the data of interest based on each data value of the element;
Based on the calculated input stimulus value, a means for updating the data value of the data of interest, which is determined according to the current data value related to the data of interest, and a different threshold for each data value; A data value update unit that compares the input stimulus value to determine whether to change the data value, and updates the data value when it is determined to change;
Including
The means for updating the data value is:
When the input stimulus value is 0 to X1min, it is Ymin, continuously increases monotonically from X1min to X1max, and when X1max or more, the first monotonically increasing function f1 becomes Ymax, and the input stimulus value is from X2min to X2max And a second monotonically increasing function f2 that continuously increases monotonically during the period, and becomes Ymax above X2max,
When the input stimulus value for the data of interest is between X1min and X2max,
Regarding the data value Y of the attention data before update, referring to I2 where Y = f2 (I2), if the input stimulus value I exceeds the I2, the data value is updated to f2 (I),
Regarding the data value Y of the attention data before update, referring to I1 where Y = f1 (I1), when the input stimulus value I is less than this I1, the data value is updated to f1 (I),
Based on the data value for at least one of the data elements after repeatedly performing the calculation of the input stimulus value and updating the data value until the values of the plurality of data elements no longer change , A data analysis apparatus characterized in that a relative distance between each data element is determined, and graphic elements corresponding to each data element are arranged and displayed so as to be separated by the determined relative distance.

Images