JP2001013864A - Device and method for correcting examination paper and recording medium recording program for the same thereon - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to correct examination papers according to answers of learners by generating correction results based on an answer corresponding correct answer tree generated by an answer corresponding correct answer tree generating means, and displaying the correction results generated by a correction result generating means. SOLUTION: The device comprises a formula database 3, a question presenting means 4 for presenting questions to a learner 1, an answer input means 5 for inputting answers, an answer corresponding correct answer tree generating means 7 for generating an answer corresponding correct answer tree including each answer formula and numerical expressions thereto in nodes, a correction result generating means 13 for generating the correction results based on the answer corresponding correct answer tree generated by the answer corresponding correct answer tree generating means 7, etc. In such a configuration, the answer corresponding correct answer tree generating means 7 evaluates each answer numerical expression, and generates an answer corresponding correct answer tree including each answer expression and formula expressions to each answer expression in nodes. Following this, the correction result generating means 13 generates the correction results based on the answer corresponding correct answer tree generated by the means 7 and displays the correction results to the learner 1 by a correction result display means 14.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an answer correction device, a method thereof, and a recording medium storing the program.

[0002]

2. Description of the Related Art In recent years, problems that can be solved by applying formulas are
2. Description of the Related Art An answer correction device for presenting to a learner, allowing the learner to input an answer composed of mathematical expressions, and correcting the answer has been put into practical use. In this type of device, only the correct answer value is obtained,
In many cases, the correct answer value is compared with the answer value to point out the inconsistency, and a crest-type idea is adopted.

[0003] However, in this case, even if the learner is making a detour, even if he enters an answer that can lead to a correct answer, the learner will unilaterally make an error, which distracts the learner's willingness to learn. Tends to.

[0004]

SUMMARY OF THE INVENTION Accordingly, the present inventors
Earlier, a proposal for improving this point was made (Japanese Unexamined Patent Publication No.
-160185). In this method, an answer tree is generated from an answer, and a correct answer tree is generated from a question and a formula. According to this, correction can be made by comparing the answer tree with the correct answer tree closest to the answer tree, which is advantageous in that correction can be performed along the answer.

[0005] Now, the "problem" to be addressed in this type of device
Presupposes that the learner knows some formulas and questions their knowledge and understanding of these formulas. And this kind of problem expects the learner to apply the parameters given in the problem to an appropriate formula and to calculate the required parameters by calculation. Therefore, the "answer" is a formula that the learner deems appropriate,
It is a group of formulas (referred to as "response formulas" in this specification) in which values corresponding to the parameters are substituted.

However, in such a problem, there is no guarantee that the procedure to reach the correct answer is one. Rather, it is normal that there is another solution, and furthermore,
It is not uncommon for there to be tens or hundreds of different solutions.

However, in the prior art proposed above, in order to perform correction in accordance with a student's answer, prior to evaluating the answer, all correct answer trees that can lead to the correct answer are obtained by the number of different answers. . Then, an answer tree is generated from the answer, and the similarity is obtained for each node of the answer tree and the correct answer tree group, and the correct answer tree closest to the answer tree is searched.

FIG. 12 schematically shows these trees. FIG. 12A shows the answer tree, and FIG.
(B) shows a group of answer trees. As shown in the figure, when the number of different solutions is large, the number of correct answer trees is also large. Then, since all of the many correct answer trees are compared with the answer tree, the amount of calculation becomes enormous.

Therefore, the processing time becomes very long,
After entering the answer, the learner may have to wait a long time to get the correction result, forgetting what he thought when entering the answer, or having to wait for distractions. However, there is a problem that the learning motivation is easily reduced. Further, since the load on the system increases, there is also a problem that it is difficult to implement the system only with a relatively inferior processing capacity such as a client machine.

Therefore, the present invention relates to an answer correction apparatus, a method thereof, and a method thereof, which not only can perform correction in accordance with a student's answer, but also can shorten the processing time and can present a highly responsive correction result to the learner. It is intended to provide a recording medium.

[0011]

According to the present invention, a problem presenting means presents a problem to a learner. The answer input means allows the learner to input an answer consisting of one or more answer formulas for the presented problem. The answer storage means stores information on the input answer. The answer-corresponding correct answer tree generating means evaluates each answer formula indicated by the answer information stored in the answer storage means, and generates an answer-corresponding correct answer tree including each answer formula and an official formula for this answer formula in a node. . The correction result generating means generates a correction result based on the answer-corresponding correct answer tree generated by the answer-corresponding correct answer tree generating means. Then, the correction result display means displays the correction result generated by the correction result generation means to the learner.

Thus, not only can the correction be made in accordance with the student's answer, but also the processing time can be reduced, and the correction result with high responsiveness can be presented to the learner.

[0013]

In the answer correcting apparatus according to the first aspect of the present invention, there is provided an official database in which official information is accumulated, and a question presenting means for presenting to the learner a question which can obtain a correct answer by a formula in the official database. Answer input means for the learner to input an answer consisting of one or more answer formulas for the question, an answer storage means for storing information of the input answer, and an answer storage means for storing the answer stored in the answer storage means. Each answer formula indicated by the information is evaluated, and each answer formula and an official formula for this answer formula are generated. A correction result generating means for generating a correction result based on the corrected answer tree corresponding to the answer, and a correction result display means for displaying a correction result generated by the correction result generating means to a learner.

With this configuration, only one answer-corresponding correct answer tree is generated in principle along with the answer, and the amount of comparison can be greatly reduced even when there are many different answers. For this reason, after the learner inputs the answer, the correction result can be immediately presented to the learner before the learner's memory or motivation is lost, and the learning effect can be improved. In addition, since the processing load can be reduced, the system can be easily implemented even in a system having a low processing capacity.

Needless to say, a technology for handling information using elements corresponding to nodes even if the tree itself does not have the apparent structure is included in the above configuration.

In the answer correction device according to the second aspect, the number of nodes is increased so that the answer-corresponding correct answer tree grows sequentially corresponding to each row of the answer formula.

With this configuration, the correct answer tree is not generated in a number of ways separately from the answer, but is generated sequentially along each line of the answer formula, so that the correction can be made faithfully to the answer.

According to a third aspect of the present invention, there is provided an answer correcting apparatus, comprising: a similarity calculating means for calculating a similarity between the answer formula and the official formula. A possible formula candidate is extracted, and a formula formula related to a formula candidate having the highest similarity with the answer formula is included in the node.

With this configuration, the closest formula can be made to correspond to the answer formula by the similarity.
Correction according to the learner's intention can be performed.

In the answer correction apparatus according to the fourth aspect, the answer-corresponding correct answer tree generating means determines that a solution has been reached if there is a parameter required for the question on the right side of the answer equation, and sets the problem answer on the right side of the answer equation. If there is no parameter to be found, it is determined that it is a dead end unless the answer is blank.

With this configuration, it is possible to clearly determine whether or not the answer has been resolved, and even if the learner has not been able to reach a solution, at least partially corrective guidance can be performed.

In the answer correction device according to the fifth aspect, the correction result generating means is partially correct as far as the answer formula is concerned, if each answer formula and an official formula for the answer formula match. If they do not agree,
In addition to pointing out that the answer formula is partially incorrect as far as the answer formula is concerned, the correction result generation means determines that the application of the formula is correct based on the answer corresponding to the answer formula that specifies that the answer formula is partially incorrect. He points out that the idea itself is correct.

According to this configuration, not only can the instruction be performed while discriminating between the correct part and the erroneous part, but also the correction result reflecting the learner's understanding as much as possible can be displayed, and the learning effect can be further improved.

According to a sixth aspect of the present invention, there is provided an answer correcting apparatus comprising an optimum solution generating means for obtaining an optimum solution from a parameter given to a problem to a parameter required for the problem, wherein the correcting result generating means is provided when the answer is blank. , And generates a correction result indicating the optimal solution.

With this configuration, when the learner cannot understand at all, the optimum solution can be taught.

Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of an answer correction device according to an embodiment of the present invention. As shown in the figure, the answer correction device is for allowing the learner 1 to solve a problem and to correct the answer, and includes the following elements.

First, the question storage means 2 is composed of, for example, a file and stores information on the question.
Alternatively, a question may be generated each time a question is asked without storing the question in advance. The official database 3 stores a plurality of official information. In this case, the questions set in the information processing engineer test are dealt with, and the following formula is assumed to be in the official database 3. Formula 1: TB * ST = SB Formula 2: SB * DS = DB Formula 3: TR * ST = SR Formula 4: SR * DS = DR Formula 5: TB * BK = TR Formula 6: SB * BK = SR Formula 7 : DB * BK = DR formula 8: TR / BK = TB formula 9: SR / BK = SB formula 10: DR / BK = DB In addition, BK (blocks) appear on the left or right side of these formulas as parameters. Conversion factor), DB (the number of blocks that can be recorded on the disk), DR (the number of records that can be recorded on the disk), DS (the number of cylinders on the disk), SB (the number of blocks that can be recorded on the cylinder), SR
(The number of records that can be recorded on a cylinder), ST (the number of tracks in a cylinder), TB (the number of blocks that can be recorded on a track), and TR (the number of records that can be recorded on a track)
One.

In FIG. 1, the answer list 6 is an example of answer storage means for storing the entered answer. Immediately after the answer is entered, the character string (answer) of each line of the multi-line edit control 5a shown in FIG. Equation) is stored.

The answer-corresponding correct answer tree generating means 7 performs processing in accordance with the flowcharts of FIGS. 4 and 5, evaluates each answer formula, and stores each answer formula and an official formula for the answer formula in a node. Generate a correct answer tree that includes the answer. The details will be described later.

When the answer formula and the formal formula are given from the answer-corresponding correct answer tree generating means 7, the similarity calculating means 8 calculates the similarity based on the following equation and returns it to the answer-corresponding correct answer tree generating means 7. . Similarity = 2α−β−γ However, all numerical values and all operators (except “=”) and a set of right-side values included in the left-hand side of the formula are set A, and are included in the left-hand side of the answer formula. When all numerical values and all operators (except for “=”) and a set of right-side values are set as set B, α: the number of elements included in both set A and set B, β: included in set A Γ: the number of elements that are included in the set B and are not included in the set A.

The formula list 9 holds a group of applicable formulas among the formulas stored in the formula database 3. Here, if there is a change in the type of the parameter that is known (in the known node list as described later),
The number of formulas listed in formula list 9 also increases or decreases.

The official candidate list 10 temporarily stores a group of formulas of interest, and a detailed role will be described later.

The known node list 11 is a list of nodes relating to currently known parameters. Immediately after inputting a response, only nodes relating to parameters given in the question exist. Then, when the formula is applied and a new parameter is known, a node relating to the parameter that has become known is added to the known node list 11 under a certain condition. If the parameter relating to the added node is a parameter required by the problem, it means that the solution has been reached.

The reserve node list 12 is for storing a reserve military node which is a path which is not growing by increasing the number of nodes but may grow in the future.

Next, the relationship between the above-mentioned nodes and the answer-corresponding correct answer tree will be described in detail with reference to FIGS.
In this example, nodes that may exist in the known node list 11 or the spare node list 12 have a data structure shown in FIG. That is, a node-type variable (typically,
A structure or an object) is prepared below a certain start address in a storage area (which may be a memory, a disk device, a stream, or the like), and the start address is indicated by a node type pointer P.

The node type variable has the following fields. That is, a node number uniquely determined for each node, an identifier assigned to a parameter in the formula (for example, BK for a blocking factor), a answer value for inserting the right-hand side value of the answer formula, and an official formula associated with the answer formula The official value to insert the right-hand side value, the similarity to insert the value obtained by the similarity calculation means 8 for the pair of the answer formula and the official formula (however, initially, when this node has children, if this node has children, The similarity and the sum of the values obtained by the similarity calculation means 8), the child node on the left side of the node (the node immediately below)
Is the left child number for entering the node number of the right child node, the right child number for entering the node number of the right child node, the answer formula that contains the answer formula itself, and the official formula that contains the official formula itself corresponding to this answer formula.

In this example, as described in Formulas 1 to 10, only two fields including left and right child numbers are prepared since the left side includes two parameters. Furthermore, in order to support a formula having three or more parameters on the left side, a pointer to a list of child numbers may be inserted. Thus, the “formula” in the present invention is not limited by the number of parameters on the left side. Further, in this example, the “formula” has only operators for performing the four arithmetic operations and operators for exponentiation, but may include mathematical functions such as trigonometric functions and exponents / logarithms.

The "node" according to the present invention is included in the "node" as long as it is associated with a value as described above, and is not necessarily limited to the data structure shown in the figure. Not necessarily.

Then, the value of each field can be set or referred to as follows. For example, FIG.
As shown in (a), to set the value “10” in the answer value field of the node pointed by the pointer P, “P−> answer value =
10; ". To substitute the variable Val with reference to the similarity of the node indicated by the pointer P, use “Val
= P->similarity;". Of course, using an object-oriented language (such as C ++ or Java), a method for setting or referencing a value is prepared, only calls of these methods are permitted, and the value of each field is hidden from the outside. You can ban any direct access. As described above, the access method to the “node” in the present invention can be arbitrarily selected.

As briefly mentioned before, as shown in FIG. 2B, the known node list 11
You can move a node to Here, FIG.
A pointer of the same type as the pointer P of (a) is stored in the known node list 11 (PA, PB,.
(PX, PY). Of course, these pointers PA, PB,..., PX, PY are not invalid pointers but point to substantial node type variables.

Now, it is assumed that the pointer PY in the spare node list 12 has been moved to the known node list 11. Then, one element is reduced from the spare node list 12 and one element is increased from the known node list 11. As will be described later, the next path (that is, a new node) from the node related to the pointer of the known node list 11 is to be grown. Therefore, when the pointer of the known node list 11 increases, a new node is generated. , It may be the solution (the parameter the problem is seeking). In other words, the node pointed to by the pointer in the spare node list 12 is nothing but a node stocked for future tree growth until the solution is reached. This movement is performed for the pointer related to the node having the highest similarity. Note that these lists 11 and 12 may be empty.

As described above, in this example, a node is configured to include the left and right child numbers. Therefore, FIG.
When attention is paid to a certain node Node3 indicated by a black circle in (a), the field of the left child number of this node Node3 is the number of another node Node1, and
If the field of the right child number of this node Node3 is the number of another node Node2, FIG.
A partial tree as shown in FIG. Further, it will be understood that if such a generation of the partial tree is repeated, the partial tree grows and a tree as shown in FIG. 3B can be formed.

The tree configured in this way is a solution-correspondence answer tree, and the solution-correspondence answer tree generating means 7 generates a solution-correspondence answer tree by operating nodes. As described above, when a pointer indicating a child number list is used instead of the left and right child numbers, a partial tree (here, an example with three children) as shown in FIG. 3C can be configured. Thus, 3
With a subtree with more than one child, as above,
The answer tree corresponding to the answer can be generated.

In the above description, the known node list 11 and the spare node list 12 include pointers indicating node type variables. This way,
With only a small pointer operation, these lists 11, 1
2 can be performed, and the processing time can be reduced, which is preferable. However, these lists 11, 1
2 may contain the node type variable itself. further,
Instead of these lists 11 and 12, an array or table having a sufficient capacity may be used instead.

Now, referring to FIG. 1, the correction result generating means 1
3 generates a correction result based on the answer-corresponding correct answer tree generated by the answer-corresponding correct answer tree generating means 7, and the corrected result is displayed to the learner 1 by the correction result display means 14 as shown in FIG. You.

Next, referring to FIGS. 4 and 5, the process of generating a correct answer tree corresponding to a question will be described with reference to FIGS. explain. This process is a process of the answer-corresponding correct answer tree generating means 7.

First, in step 1 in FIG. 4, a timer which times out after a predetermined time is set. This timer is for avoiding an infinite loop, and may be replaced by a loop counter or the like, or may be omitted if there is no possibility of an infinite loop.

Next, known parameters are extracted from the question sentence that is set by the question storage means 2 or generated each time the question is set, and some node-type variables corresponding one-to-one with the known parameters are extracted. Generated and pointers to these variables are set in the known node list 11. At this time, each node is assigned a unique (non-overlapping) node number. Here, these nodes are the lowest-layer (having no children) nodes in FIG. 3B, and in this state, the given identifier is entered in the identifier field, and the value is entered in the official value field. Enters. The left child number, the right child number, the answer formula, and the formula formula each contain an initial value indicating invalidity, and the similarity field contains "0". Then, the official value is directly substituted into the answer value field. In other words, if any of the left child number, right child number, answer formula, formula formula fields are invalid initial values, this node corresponds to the parameter given in question,
This means that the bottom layer has no child nodes.

In the example of FIG. 10, the parameter to be obtained is D
B, TB = 20 (node number: 1), BK = 10
(Node number: 2), ST = 30 (node number: 3),
Four new node type variables of DS = 80 (node number: 4) are newly generated, and pointers to these are set in the known node list 11.

Next, in step 3, the spare node list 12 is emptied.

Next, at step 4, it is checked whether the answer set in the answer list 6 from the answer input means 5 is not blank. If it is blank, there is no way to generate an answer-corresponding correct answer tree, so a blank page determination (step 5) is made and the process ends.

Here, as shown in FIG. 10, since the answer is not a blank answer, the process proceeds to step 6. At step 6, the answer list 6 and the official list 9 are set. At this point, the answer list 6 includes all lines of the answer formula (4 in FIG. 10).
Line) enters. In the official list 9, all the officials of the official database 3 (here, officials 1 to 10) are set.

Next, in step 7, formulas having parameters existing in the known node list 11 on the right side are excluded from the formula list 9. This excludes formulas that do not generate new nodes (that is, they do not approach the solution). Here, formula 8 having TB on the right side is removed, and formula list 9 has 9 formulas.

Next, it is checked whether the official list 9 is not empty (step 8). If it is empty, it is interpreted that there is no longer any applicable formula and that it is in a jammed state, a deadlock is determined (step 10), and the process ends. In step 9, it is checked whether the time is up. If so, it is interpreted that there is a high possibility that an infinite loop has occurred, and the process is similarly terminated.

Here, since the official list 9 is not empty, the process proceeds to step 11.

In step 11, the known node list 11
Check if there are any parameters you want. If so, it is determined that a solution has been reached (step 12), and the process ends. Here, the known node list 11 includes the parameter D
Since there is nothing related to B, the process proceeds to step S13.

In step 13, node processing is performed as shown in FIG. First, the official candidate list 10 is emptied (step 21), and applicable formulas are set to the official candidate list 1
It is set to 0 (step 22). The applicable formula is a formula that can satisfy all the parameters on the left side of the formulas in the formula list 9 with the parameters of the nodes in the known node list 11.

Here, among formulas 1 to 7, 9 to 10,
The only applicable formulas are Formula 1 and Formula 5. Therefore, formulas 1 and 5 are set in the formula candidate list 10.

Next, a value is substituted for each formula candidate to obtain a right side value. For each of the formulas obtained in this way, the answer line having the maximum similarity is obtained using the similarity calculating means 8 (step 23). Here, since formulas 1 and 5 are candidates, these formulas and each answer line are examined. For Formula 1, the formula that sets a value for each parameter is “20 * 30 = 600”. The answer is as shown in FIG. 10. The answer formula first line: common term {20, *}, similarity 2 * 2-
2-2 = 0 answer formula second line: common term {*, 30}, similarity 2 * 2-
2-2 = 0 Response formula line 3: common term None, similarity 2 * 0-3-3
= -6 The answer formula 4th line: No common term, similarity 2 * 0-3-3
= −6, and the highest similarity is in the first and second rows.
For Formula 5, the formula that sets a value for each parameter is “20 * 10 = 200”. The answer is Figure 10
The first line of the answer formula: common terms {20, *, 10, 200},
Similarity 2 * 4-0-0 = 8 Second line of answer formula: common term {*}, similarity 2 * 2-2-2
= 0 The third line of the answer formula: No common term, similarity 2 * 0-3-3
= -6 The answer formula 4th line: common term {10}, similarity 2 * 1-3-
3 = −4, and the highest similarity is in the first row.

In the above description, the maximum similarity is the formula 5 of formula 5 “20 * 10 = 200” and the answer formula 1st line.

Next, in step 24, a node (pointer) relating to the line of the obtained answer formula is set in the spare node list 12, each value is set, and the corresponding line is deleted from the answer list 6.

Here, the first line of the answer formula is deleted from the answer list 6, and a new node type pointer is added to the spare node list 12.
The number of 2 becomes 1. Then, a unique node number (here, 5) is assigned to the node type variable pointed to by the pointer, the identifier is “TR” (identifier of the right side parameter of the formula 5), and the answer value is “200” (answer formula). The first line's right side value), the official value is "200" (the right side value of the formula), the similarity is "8", the left child number is "1" (the node number related to the parameter TB), and the right child number is " 2 ”(node number related to parameter BK), and“ 20 * 10 = 20 ”
0 "and" 20 * 10 = 200 "are set in the formula.

Here, the value “8” set in the similarity field is the value “8” obtained by the similarity calculating means 8.
And the similarity “0” of the left child number node and the similarity “0” of the right child number node.

Next, in step 25, the node having the highest similarity in the spare node list 12 is moved to the known node list 11. Therefore, the node with the highest credibility has a chance to move to the known node list 11 and receive the next formula application (used as a parameter on the right side). At the same time, other nodes remaining in the spare node list 12 are also stored afterwards so that this chance can be seen.

Here, the spare node list 12 contains 1
Since there are only two nodes, the node with the node number 5 described above moves to the known node list 11.

Next, the process proceeds to step 14 in FIG. 4, and it is checked whether the answer list 6 is empty. If it is empty, it is examined in step 15 whether a node related to the parameter to be found exists in the known node list 11. If it exists, it is determined that the solution has been reached (step 12), and the processing ends. If it does not exist, the answer is determined to be a dead end, assuming that the answer has not reached the solution (step 10), and ends.

On the other hand, if it is not empty, the processing from step 7 onward is repeated for the remaining lines of the answer list 6.

Here, the process finally proceeds from step 15 to step 12, where it is determined that a solution has been reached.

With the above processing, an answer-corresponding correct answer tree is generated unless the answer is a blank answer. Here, such processing is repeated, and as the number of generated nodes increases,
The answer-corresponding correct answer tree has a deep hierarchy and a high tree structure. This answer-correspondence correct answer tree has the same hierarchy depth and the same structure as the answer tree that is composed of only answers.

From the above description, it can be understood that the answer-corresponding correct answer tree is usually one tree having the same structure as the answer tree. That is, FIG. 9 (the present invention) and FIG.
As is clear from comparison with (prior art), in the present invention, the number of objects to be compared is significantly reduced in comparison with the conventional art.

Next, a process of generating a correction result by the correction result generating means 13 will be described with reference to FIGS.
First, if a blank page is determined in step 31, the optimum solution is generated by the optimum solution generating means 15 (step 32), and the correction result display means 14 is displayed by the optimum solution (step 37). .

If the deadlock is determined (step 3
3) The correct answer tree corresponding to the answer (although halfway) is reproduced from the known node list 11 to evaluate each node related to the answer input by the learner 1 (step 34), and the result of the evaluation is added to the correction result. reflect. Further, the answer input by the learner 1 is deduced, and an attempt is made to automatically create an answer leading to a correct answer (step 35).

Further, if it is determined that the solution has been reached (steps 31 and 33), the answer corresponding correct answer tree (completed to the solution) is reproduced from the known node list 11 and each node is evaluated (step 36). ), And displays the correction result along with the answer (step 37).

Next, the above “evaluation” will be described in detail with reference to FIG. Here, “evaluation” in the present example refers to FIG.
Steps 34 and 36 are performed to determine whether each line of the answer formula is correct (correct), whether the concept is correct but the result is different (quasi-correct), or completely wrong (incorrect). Things. Specifically, the known node list 11
Based on each node in

First, in step 41, the node of interest is set as the first node of the known node list 11. Next, it is checked whether or not the evaluation has been completed for all nodes (step 42), and if not, it is checked whether or not the target node is the lowest layer (has no child nodes) (step 4).
3). As described above, if the left child number or the like of the node of interest is an initial value indicating invalidity, it is the lowest layer, otherwise it is not the lowest layer, so this check can be easily performed. Also, the reason for performing this check and excluding the lowest-level node is that the parameters (given in the question) relating to the lowest-level node are not obtained by the learner 1 but are included in this “evaluation”. This is because it does not fit.

If the node is the lowest layer, at step 50, the node of interest is advanced by one without evaluation and the process returns to step 42. Otherwise, in step 44, it is examined whether the formula of the target node matches the answer formula. If they match, it can be said that the learner 1 has input a completely correct answer formula for this node of interest, and a positive evaluation is performed (step 45).

If the results of the examination at step 44 do not match, it is examined whether or not the idea is correct. Here, it has been confirmed in step 43 that the target node is not the lowest layer, and the target node always has a child node. Then, by referring to the right child number and left child number of the node of interest, the answer value of the right child node and the answer value of the left child node are obtained, and these answer values are substituted into the corresponding formula, and the right side value is obtained. Is temporarily stored as a reference value (step 46). Then, it is examined whether or not the reference value matches the answer value of the target node (step 47). If they match, the concept is deemed to be correct, and a quasi-evaluation is performed (step 48). If they do not match, an incorrect evaluation is performed as a complete error (step 49). Then, the target node is advanced by one (step 50), and the process returns to step 42.

Here, if the learner 1 makes a mistake at a certain stage of solving the problem, the resulting right-side value becomes invalid. Then, if this incorrect right-hand side value is used at a later stage, only an incorrect value can be obtained even if the concept of applying the formula is correct. However, by performing the above processing, the learner 1 can be rescued even in such a case,
It is possible to improve learning motivation.

Next, the above “deduction” will be described in detail with reference to FIG. Here, "deduction" is a process of searching for a solution that proceeds from a solution that has not yet been solved to a solution.
Specifically, the following is performed.

As shown in FIG. 8, this “deduction”
Can be realized by a process similar to the flowchart of FIG. First, at step 51, a timer that times up after a predetermined time is set to avoid an infinite loop. Next, the known node list 11 is left as it is, and the spare node list 12
And set the official list 9 (step 52,
53), remove meaningless formulas from the formula list 9 (step 54).

Then, it is checked that the official list 9 is not empty and not jammed, and that the time has not expired (steps 55 and 56). If this check result is negative, it is determined that the solution cannot be reached even if the answer is advanced, and an optimal solution is generated (step 60), and the fact that the optimal solution and the solution are inappropriate is added to the display contents (step 6).
1).

If the check result is correct, the parameters required in the known node list 11 are found (step 5).
7) A new node is generated and added to the known node list 11 (step 58). Then, when the required parameter is reached, the deduction result is added to the display content (step 59). At this time, it is conceivable that the solution may be a detoured solution compared to the optimal solution, but it is preferable because the guidance according to the thinking of the learner 1 can be performed. In this "deduction", the answer itself does not exist,
The similarity calculation as shown in FIG. 5 is not performed.

Now, in the example shown in FIG. 10, the first and second
Line 3 is "positively evaluated", line 3 is "illegally evaluated"
The fourth line is “quasi-positive evaluation”. It is also determined that a solution has been reached.

For this reason, the correction result display means 14 is provided in FIG.
As shown in 1, the correction result is displayed. As is clear from FIG. 11, correct / incorrect is pointed out for each line of the answer formula based on the above “correct evaluation”, “incorrect evaluation”, and “quasi-right evaluation”. Instructs the law. Furthermore, the fourth line points out that although the concept is based on the "quasi-correct evaluation", the result is different. Such an indication can be easily realized because the answer value and the official value are held in association with each other.

As shown in the figure, displaying a reference message is more preferable because the learning effect can be further enhanced.

Here, the "recording medium in which the answer correction program is recorded in a computer-readable manner" as referred to in this specification.
Includes a case where the program is distributed and distributed to a plurality of recording media. Also, regardless of whether this program is a part of the operating system or not, various processes or threads (DLL, OCX, ActiveX, etc. (including the trademark of Microsoft Corporation)) can take over some of the functions. In this case, the part related to the replaced function is not stored in the recording medium.

Further, the problem presentation means 4 and the like may be replaced by an HTML file that can be displayed on a WWW browser, and the answer-corresponding correct answer tree generating means 7 may be constituted by an applet / application cited in the HTML file. In this case, the WWW browser is an entity of the problem presenting means 4, and only the data displayed on the entity actually exists on the recording medium. The HTML file and the applet / application are usually located on a server and are basically distributed to the client via a network. However, when the HTML file and the applet / application are located on the server, the HTML file and the applet / application are stored in the server. The storage device is the “recording medium” here,
When it is on the client, the recording device of the client is the “recording medium” here.

Further, in the "program", in addition to the application compiled into the machine language, the case where the program actually exists as an intermediate code interpreted by the above-described process or thread, or at least the resource and the source code are "recorded". If a compiler and linker that can generate a machine language application from them are stored on the recording medium, at least resources and source code are stored on the recording medium, and intermediate code This includes the case where an interpreter capable of generating an application is present in the “recording medium”.

In addition to the case where only one terminal includes all the elements appearing in the present specification, one terminal is a client and is connected to a connectable server or network. All or some of the elements may exist. Typically, a form is conceivable in which the server has the official database 3, the problem storage unit 2 or the unit for generating the problem each time, and the optimal solution generation unit 15 and the client has other elements.

[0090]

According to the first, seventh, and thirteenth aspects of the present invention, even when there are many different solutions, the answer-corresponding correct answer tree is:
According to the answer, only one method is generated in principle,
Compared to the conventional technology, the amount of comparison and the amount of calculation can be significantly reduced,
Not only can the correction be made according to the student's answer, but also the processing time can be shortened and the correction result with high responsiveness can be presented to the learner. Further, even in an environment where the processing capacity is relatively low, it can be operated with practicality.

According to the second, eighth and fourteenth aspects of the present invention, since the number of nodes is increased so as to grow sequentially corresponding to each row of the answer formula, a correct answer tree faithfully reflecting the answer is generated, and the sequential correction is performed. I can do it.

According to the third, ninth and fifteenth aspects of the present invention, based on the degree of similarity, a correction can be made that faithfully reflects the intention of the learner.

According to the fourth, tenth, and sixteenth aspects of the present invention, the answer that has reached the solution is corrected according to the answer, and even if the solution cannot be reached, it can be dealt with accordingly.

According to the fifth, eleventh, and seventeenth aspects of the present invention, not only correct / incorrect instruction, but also the part that the learner can understand can be evaluated, so that the learning effect can be further improved.

According to the sixth, twelfth and eighteenth aspects, it is possible to teach an optimal solution to a learner who has no understanding at all.

[Brief description of the drawings]

FIG. 1 is a block diagram of an answer correction device according to an embodiment of the present invention.

FIG. 2A is a data structure diagram of the same node; FIG. 2B is an explanatory diagram of node movement from the spare node list to a known node list;

3A is a diagram illustrating a partial tree reproduced from the same node; FIG. 3B is a diagram illustrating a correct answer tree corresponding to an answer reproduced from the same node; FIG. 3C is a diagram illustrating a partial tree reproduced from the same node;

FIG. 4 is a flowchart illustrating a processing process according to the answer tree corresponding to the answer tree;

FIG. 5 is a flowchart exemplifying a processing process according to the answer plan corresponding correct answer tree generating means;

FIG. 6 is a flowchart illustrating a processing process according to the correction result generation unit;

FIG. 7 is a flowchart illustrating the evaluation process.

FIG. 8 is a flowchart illustrating the deduction process.

FIG. 9A shows an example of the answer tree. FIG. 9B shows an example of the answer tree corresponding to the answer.

FIG. 10 is a diagram showing an example of display by the question presenting means and the answer input means.

FIG. 11 is a diagram showing a display example by the correction result display means.

12A is a diagram illustrating a conventional answer tree, and FIG. 12B is a diagram illustrating a conventional correct answer group.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 learner 2 question storing means 3 official database 4 question presenting means 5 answer input means 6 answer list 7 answer corresponding correct answer tree generating means 8 similarity calculating means 9 official list 10 official candidate list 11 known node list 12 spare node list 13 correction Result generation means 14 Correction result display means 15 Optimal solution generation means

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Yuichi Kawaguchi 1006 Kazuma Kadoma, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. F term (reference) 2C028 AA12 BB04 BC01 BC02 BD02 CA12 DA07

Claims (18)

[Claims] 1. An official database in which official information is stored; a question presenting means for presenting a learner with a correct answer by a formula in the official database; Answer input means for inputting an answer consisting of a plurality of answer formulas, answer storage means for storing information on the input answer, and evaluating each answer formula indicated by the answer information stored by the answer storage means, Answer answer tree generating means for generating an answer answer tree containing nodes, each answer equation and an official equation for this answer equation, and correction based on the answer answer tree generated by the answer answer tree generating means. An answer correction device comprising: a correction result generating means for generating a result; and a correction result display means for displaying a correction result generated by the correction result generating means to a learner. 2. The answer correcting apparatus according to claim 1, wherein said answer-corresponding correct answer tree generating means increases the number of nodes so as to grow sequentially corresponding to each row of the answer formula. 3. A similarity calculating means for calculating a similarity between the answer formula and an official formula, wherein the answer-corresponding correct answer tree generating means extracts an applicable formula candidate from a formula group of the formula database. 3. The answer correction device according to claim 1, wherein an official formula related to an official candidate having the highest similarity with the answer formula is included in the node. 4. The answer-corresponding correct answer tree generating means determines that a solution has been reached when there is a parameter required by the question on the right side of the answer formula, and determines that the parameter required by the question does not exist on the right side of the answer formula. 4. The answer correcting apparatus according to claim 1, wherein the apparatus determines a dead end unless the answer is a blank answer. 5. The correction result generating means includes: each answer formula;
If the formula for this answer formula matches,
It points out that the answer formula is partially correct as far as this answer formula is concerned, and if it does not match, it points out that it is partially incorrect as far as this answer formula is concerned, and the correction result generation means specifies that it is not partially correct. 5. The answer correction device according to claim 1, wherein when the answer formula to be executed can determine that the application of the formula is correct based on the answer corresponding correct answer tree, the idea itself is pointed out to be correct. 6. An optimum solution generating means for obtaining an optimum solution from a parameter given to the problem to a parameter required by the problem, wherein the correction result generating means, when the answer is blank, indicates a correction result indicating the optimum solution. The answer correction device according to claim 1, wherein the answer correction device generates the answer. 7. A step of presenting, to a learner, a question that can be correctly answered by a formula stored in an official database, and providing one or a plurality of answer formulas input by the learner for the presented question. Receiving the input, storing the entered answer information, evaluating each answer formula indicated by the stored answer information, and including each answer formula and an official formula for this answer formula in a node. A method for correcting an answer, comprising: a step of generating a corresponding answer tree; a step of generating a correction result based on the answer corresponding tree; and a step of displaying the correction result to a learner. 8. The method according to claim 7, wherein the number of nodes is increased so as to grow sequentially in accordance with each line of the answer formula. 9. The method according to claim 1, further comprising the step of determining a similarity between the answer formula and the formula, extracting applicable candidates from the formula group of the official database, and formulating the formula candidate having the highest similarity with the answer formula. 9. The method according to claim 7, wherein the formula is included in a node. 10. When there is a parameter required by the question on the right side of the answer formula, it is determined that a solution has been reached. When no parameter required by the problem exists on the right side of the answer formula, it is determined that the answer is a dead end unless the answer is blank. The method according to claim 7, wherein the answer is corrected. 11. If each answer formula and the formula for this answer formula match, it is pointed out that the answer formula is partially correct as far as this answer formula is concerned,
If it is pointed out that the answer formula is not partially correct as far as this answer formula is concerned, and the answer formula that specifies that it is partially incorrect can be determined that the application of the formula is correct based on the answer corresponding answer tree,
11. The method according to claim 7, wherein the idea is pointed out to be correct. 12. A method for determining an optimal solution from parameters given to a problem to parameters required by the problem,
12. The method according to claim 7, wherein when the answer is blank, a correction result indicating an optimal solution is generated. 13. A problem presenting means for presenting to a learner a problem that can be correctly answered by a formula stored in an official database, and the learner responding to the presented problem with one or more answer formulas. Answer input means for inputting, answer storage means for storing information on the input answer, evaluating each answer formula indicated by the answer information stored in the answer storage means, each answer formula, and this answer formula And a correct answer tree generating means for generating a correct answer tree corresponding to the node, and a correction result generating means for generating a correction result based on the correct answer tree generated by the correct answer tree generated by the correct answer tree. And a computer-readable recording medium storing an answer correction program comprising: a correction result display means for displaying a correction result generated by the correction result generation means to a learner. 14. The answer correction program according to claim 13, wherein said answer corresponding correct answer tree generating means increases the number of nodes so as to grow sequentially in accordance with each line of the answer formula. recoding media. 15. A similarity calculating means for calculating a similarity between an answer formula and an official formula, wherein the answer-corresponding correct answer tree generating means extracts an applicable formula candidate from a formula group of the formula database. 15. The computer-readable recording medium according to claim 13, wherein an official formula related to an official candidate having the highest similarity to the answer formula is included in the node. 16. The answer-corresponding correct answer tree generating means, if a parameter required by the question exists on the right side of the answer formula,
14. The method according to claim 13, wherein it is determined that a solution has been reached, and if there is no parameter required for the problem on the right side of the answer formula, a dead end is determined unless the answer is a blank answer.
A recording medium in which the answer correction program described in 5 is recorded in a computer-readable manner. 17. The correction result generation means points out that if each answer formula and an official formula for this answer formula match, it is partially correct as far as the answer formula is concerned. If not, it is pointed out that the answer formula is not partially correct as far as the answer formula is concerned, and the correction result generation means determines that the answer formula that specifies that the answer formula is partially incorrect is based on the answer corresponding correct answer tree and the application of the formula is correct. 17. The computer-readable recording medium according to claim 13, wherein when it can be determined, the idea itself is pointed out to be correct. 18. An apparatus according to claim 1, further comprising an optimum solution generating means for obtaining an optimum solution from a parameter given to the problem to a parameter required by the problem, wherein the correction result generating means includes a correction result indicating an optimum solution when the answer is blank. 18. A recording medium in which the answer correction program according to claim 13 is recorded in a computer-readable manner.