Data analyzer utilizing the spreading activation theory for stemming processing

1. A data analyzer that executes data analysis using a “Spreading Activation Theory” method, the data analyzer comprising:

a computer-readable storage unit that stores data values at any one of N stages (N is an integer not smaller than 2) while associating the data values with a plurality of data elements respectively and stores association weighting information among the data elements, the data elements being selected using a stemming process;

a selecting and computing unit that selects one of the data elements as a target data on the basis of a predetermined rule and computes an input stimulus value concerned with the target data on the basis of the association weighting information between the target data and the other data elements and the data values of the other data elements; and

an updating unit that updates the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with a current data value associated with the target data and varying according to each data value is compared with the input stimulus value to determine whether the data value needs to be changed or not, and that the data value is updated when a decision is made that the data value needs to be changed,

wherein after repeating execution of computing of the input stimulus value and updating of the data value until satisfaction of a predetermined condition, the data value concerned with at least one of the data elements is subjected to a predetermined process.

2. A data analyzer that executes data analysis using a “Spreading Activation Theory” method, the data analyzer comprising:

a computer-readable storage unit that stores data values as continuous values in a range of from Ymin to Ymax (Ymin<Ymax) while associating the data values with a plurality of data elements respectively and stores association weighting information among the data elements, the data elements being selected using a stemming process;

a selecting and computing unit that selects one of the data elements as a target data on the basis of a predetermined rule and computes an input stimulus value concerned with the target data on the basis of the association weighting information between the target data and the other data elements and the data values of the other data elements; and

an updating unit that updates the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with a current data value associated with the target data and varying according to each data value is compared with the input stimulus value to determine whether the data value needs to be changed or not, and that the data value is updated when a decision is made that the data value needs to be changed,

wherein after repeating execution of computing of the input stimulus value and updating of the data value until satisfaction of a predetermined condition, the data value concerned with at least one of the data elements is subjected to a predetermined process.

3. The data analyzer according to claim 2 , wherein

the updating unit uses a first monotonically increasing function f 1 which takes Ymin when the input stimulus value is in a range of from 0 to X 1 min, which increases continuously monotonically when the input stimulus value is in a range of from X 1 min to X 1 max and which takes Ymax when the input stimulus value is not smaller than X 1 max, and a second monotonically increasing function f 2 which increases continuously monotonically when the input stimulus value is in a range of from X 2 min to X 2 max and which takes Ymax when the input stimulus value is not smaller than X 2 max,

while the input stimulus value with respect to the target data is in a range of from X 1 min to X 2 max,

the data value updating means updates the data value to f 2 (I) by referring to I 2 satisfying Y=f 2 (I 2 ) with respect to the data value Y of the target data before updating, when the input stimulus value I is larger than I 2 , and

the data value updating means updates the data value to f 1 (I) by referring to I 1 satisfying Y=f 1 (I 1 ) with respect to the data value Y of the target data before updating, when the input stimulus value I is smaller than I 1 .

4. The data analyzer according to claim 2 , further comprising:

a scaling unit that scales data values with respect to the data elements so that the sum of the data values becomes equal to a predetermined value.

5. The data analyzer according to claim 4 , wherein

the updating unit uses a first monotonically increasing function f 1 which takes Ymin in an input stimulus value range of from 0 to X 1 min, continuously monotonically increases in an input stimulus value range of from X 1 min to X 1 max and takes Ymax in an input stimulus value range of not smaller than X 1 max, and a second monotonically increasing function f 2 which continuously monotonically increases in an input stimulus value range of from X 2 min to X 2 max and takes Ymax in an input stimulus value range of not smaller than X 2 max,

while the input stimulus value for the target data is in a range of from X 1 min to X 2 max,

the data value updating means updates the data value to f 2 (I) when the input stimulus value I is larger than I 2 by referring to I 2 to satisfy Y=f 2 (I 2 ) with respect to the data value Y of the target data before updating, and

the data value updating means updates the data value to f 1 (I) when the input stimulus value I is smaller than I 1 by referring to I 1 to satisfy Y=f 1 (I 1 ) with respect to the data value Y of the target data before updating.

6. A data analyzing method using a computer having a computer-readable storage unit that executes data analysis using a “Spreading Activation Theory” method and that stores data values as discrete or continuous values at a plurality of stages while associating the data values with a plurality of data elements respectively and stores association weighting information among the data elements, the data elements being selected using a stemming process, the method comprising the steps of:

selecting one of the data elements as a target data on the basis of a predetermined rule and computing an input stimulus value concerned with the target data on the basis of the association weighting information between the target data and the other data elements and the data values of the other data elements;

updating the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with a current data value associated with the target data and varying according to each data value is compared with the input stimulus value to determine whether the data value needs to be changed or not, and that the data value is updated when a decision is made that the data value needs to be changed; and

storing the updated data value on the computer-readable storage unit or outputting the updated data value to a display,

wherein the steps are repeatedly executed until a predetermined condition is satisfied, and

after the repeated execution, the data value concerned with at least one of the data elements is subjected to a predetermined process.

7. The data analyzing method according to claim 6 , further comprising the step of:

scaling data values with respect to the data elements so that the sum of the data values becomes equal to a predetermined value.

8. A storage medium readable by a computer, the storage medium storing a program of instructions executable by the computer having a storage unit that executes data analysis using a “Spreading Activation Theory” method and that stores data values as discrete or continuous values at a plurality of stages while associating the data values with a plurality of data elements respectively and stores association weighting information among the data elements, the data elements being selected using a stemming process, the function comprising the steps of:

selecting one of the data elements as a target data on the basis of a predetermined rule and computing an input stimulus value concerned with the target data on the basis of the association weighting information between the target data and the other data elements and the data values of the other data elements;

updating the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with a current data value associated with the target data and varying according to each data value is compared with the input stimulus value to determine whether the data value needs to be changed or not, and that the data value is updated when a decision is made that the data value needs to be changed; and

storing the updated data value on the storage medium or outputting the updated data value to a display.

9. The storage medium according to claim 8 , wherein the function further comprising the step of:

scaling data values with respect to the data elements so that the sum of the data values becomes equal to a predetermined value.

10. A data analyzer that executes data analysis using a “Spreading Activation Theory” method, the data analyzer comprising:

a computer-readable storage unit that stores data values as continuous values in a range of from Ymin to Ymax (Ymin<Ymax) while associating the data values with at least a part of data elements respectively and stores association weighting information among the data elements the data elements being selected using a stemming process;

a selecting unit that selects at least one of the data elements as a seed and selects the seed and at least one of the other data elements different from the seed as subjects of computation;

a selecting and computing unit that selects one of the selected subjects of computation as a target data on the basis of a predetermined rule and computes an input stimulus value concerned with the target data on the basis of information of association weighting between the target data and the data elements selected as the subjects of computation and the respective data values of the data elements selected as the subjects of computation; and

an updating unit that updates the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with the current data value associated with the target data and different in accordance with the data value is compared with the input stimulus value to decide whether the data value is to be changed or not, and that the data value is updated when a decision is made that the data value is to be changed,

wherein after repeated execution of computing of the input stimulus value and updating of the data value until satisfaction of a predetermined condition, the data value concerned with at least one of the data elements is subjected to a predetermined process.

11. The data analyzer according to claim 10 , further comprising:

an additionally selecting unit that additionally selects data elements as subjects of computation on the basis of the data value updated by the updating unit.

12. The data analyzer according to claim 10 , wherein

the updating unit uses a first monotonically increasing function f 1 which takes Ymin when the input stimulus value is in a range of from 0 to X 1 min, which increases continuously monotonically when the input stimulus value is in a range of from X 1 min to X 1 max and which takes Ymax when the input stimulus value is not smaller than X 1 max, and a second monotonically increasing function f 2 which increases continuously monotonically when the input stimulus value is in a range of from X 2 min to X 2 max and which takes Ymax when the input stimulus value is not smaller than X 2 max,

while the input stimulus value with respect to the target data is in a range of from X 1 min to X 2 max,

the data value updating means updates the data value to f 2 (I) by referring to I 2 satisfying Y=f 2 (I 2 ) with respect to the data value Y of the target data before updating, when the input stimulus value I is larger than I 2 , and

the data value updating means updates the data value to f 1 (I) by referring to I 1 satisfying Y=f 1 (I 1 ) with respect to the data value Y of the target data before updating, when the input stimulus value I is smaller than I 1 .

13. The data analyzer according to claim 10 , wherein the updating unit scales the data values with respect to data elements so that the sum of the data values becomes equal to a predetermined value when the data value is updated.

14. A data analyzing method using a computer having a computer-readable storage unit that executes data analysis using a “Spreading Activation Theory” method and that stores data values as continuous values in a range of from Ymin to Ymax (Ymin<Ymax) while associating the data values with at least a part of data elements respectively and stores association weighting information among the data elements, the data elements being selected using a stemming process, the method comprising the steps of:

selecting at least one of the data elements as a seed and selecting the seed and at least one of the other data elements different from the seed as subjects of computation;

selecting one of the selected subjects of computation as a target data on the basis of a predetermined rule and computing an input stimulus value concerned with the target data on the basis of information of association weighting between the target data and the data elements selected as the subjects of computation and the respective data values of the data elements selected as the subjects of computation; and

updating the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with the current data value associated with the target data and different in accordance with the data value is compared with the input stimulus value to decide whether the data value is to be changed or not, and that the data value is updated when a decision is made that the data value is to be changed,

wherein after repeated execution of computing of the input stimulus value and updating of the data value until satisfaction of a predetermined condition, the data value is output to a display or stored on the computer-readable storage unit.

15. A storage medium readable by a computer, the storage medium storing a program of instructions executable by the computer having a storage unit that executes data analysis using a “Spreading Activation Theory” method and that stores data values as continuous values in a range of from Ymin to Ymax (Ymin<Ymax) while associating the data values with at least a part of data elements respectively and stores association weighting information among the data elements, the data elements being selected using a stemming process, the function comprising the steps of:

selecting at least one of the data elements as a seed and selecting the seed and at least one of the other data elements different from the seed as subjects of computation;

selecting one of the selected subjects of computation as a target data on the basis of a predetermined rule and computing an input stimulus value concerned with the target data on the basis of association weighting information between the target data and the data elements selected as the subjects of computation and the respective data values of the data elements selected as the subjects of computation;

updating the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with the current data value associated with the target data and different in accordance with the data value is compared with the input stimulus value to decide whether the data value is to be changed or not, and that the data value is updated when a decision is made that the data value is to be changed; and

storing the updated data value on the storage medium or outputting the updated data value to a display,

wherein computing of the input stimulus value and updating of the data value are repeatedly executed until satisfaction of a predetermined condition.

16. A data analyzer that executes data analysis using a “Spreading Activation Theory” method, the data analyzer comprising:

a computer-readable storage unit that stores data values while associating the data values with at least a part of data elements respectively and stores association weighting information among the data elements;

a selecting and computing unit that selects one of the data elements as a target data on the basis of a predetermined rule and computes an input stimulus value concerned with the target data on the basis of association weighting information among the target data and the data elements selected as the subjects of computation and the respective data values of the data elements selected as the subjects of computation; and

an updating unit that updates the data value of the target data on the basis of the computed input stimulus value in such a manner that a threshold decided in accordance with the current data value associated with the target data and different in accordance with the data value is compared with the input stimulus value to decide whether the data value is to be changed or not, and that the data value is updated when a decision is made that the data value is to be changed,

wherein after repeated execution of computing of the input stimulus value and updating of the data value until satisfaction of a predetermined condition, a relative distance between data elements is decided on the basis of the data value concerned with at least one of the data elements so that graphic elements corresponding to the data elements respectively are arranged so as to be separate by the decided relative distance and displayed.