JP5806642B2 - Infant word search device, method and program thereof - Google Patents

Description

  The present invention relates to an infant word search apparatus, a method thereof, and a program for searching for words in an infant vocabulary learning process.

  In the field of studying early vocabulary development of infants, “50% reaching age” has been used as one of the indices representing the characteristics of infant words. “50% reaching age” means the number of months that multiple infants learned are observed and recorded every month, and the percentage of acquisition of each word is calculated from the cross-sectional data. That is. For example, if the word “papa” was acquired by 42% in 14 months, 51% in 15 months, and 64% in 16 months, the age at which 50% of “papa” reached 15 months would be 15 months.

Conventionally, the data of “50% reaching age” has been collected using a standardized vocabulary checklist called a MacArthur Infant Language Development Questionnaire described in Non-Patent Document 1, for example. Then, “50% reached age” was set as the feature amount of each word.
FIG. 12 exemplifies the value of “50% reaching age” of the words related to daily routines and greetings collected in this way. The first column is a word, the second column to the twelfth column are appearance rates with respect to the age of the moon, and the thirteenth column is “50% reached age”. FIG. 12 is data of about 100 infants.

Tamako Komine, Toru Watanamaki, “Standard Research on Vocabulary Development in Japanese Children: From the Japanese MacArthur Infant Language Development Questionnaire”, Development and Nursing Research Institute 2008. vol.24.

  Since “50% reaching age” used the numerical value of the month unit corresponding to the appearance rate of the word as it is, there was a deficiency in terms of resolution and accuracy as a feature amount. Also, based on the value of “50% reaching age”, for example, when trying to search for a word whose value is 15 months, conventionally, there is only a method of searching using, for example, spreadsheet software. There was a problem that the form was not general.

  The present invention has been made in view of such problems, and by introducing a function having good compatibility with the characteristics of “50% reaching age”, the resolution and accuracy of data are improved and the function is improved. It is an object of the present invention to provide an infant word search apparatus, a method and a program thereof that can easily search for a word using these parameters.

The infant word search device of the present invention includes a logistic function approximation unit, a parameter storage unit, and a word search unit. Logistic function approximation unit as input acquisition time x the cumulative utterance ratio P _w acquisition word w in a plurality of infants vocabulary learning process, to model the function of the following equation to draw a logistic curve for each acquisition word w, parameter c Calculate _w and b _w .

The parameter storage unit stores the parameters c _w and b _w . Word search unit, a word matching the question inputted, searched and calculated using the parameters c _w and b _w, and outputs a desired word.

  According to the infant word search device of the present invention, the cumulative utterance ratio characteristics of the word are modeled with a logistic curve, the cumulative utterance ratio is calculated, and the acquired word matching the input question is searched and output. It is possible to improve the resolution and accuracy as the feature amount of “50% reaching age”. In addition, it is possible to easily search for infant words.

The figure which shows the function structural example of the infant word search apparatus 100 of this invention. It shows an example of the relationship of the acquisition word w and acquisition time x in the case of "Yes" and the cumulative utterance ratio P _w. The figure which shows the operation | movement flow of the infant word search apparatus 100. FIG. The figure which shows the logistic curve obtained by the function which modeled the relationship shown in FIG. The figure which shows the logistic curve at the time of modeling a function (Formula (1)) about the word "barba". The figure which shows the function structural example of the infant word search apparatus 200 of this invention. The figure which shows the operation | movement flow of the infant word search apparatus 200. FIG. The figure which shows the logistic curve at the time of modeling a function (Formula (2)) about the word "barba". The figure which shows the logistic curve at the time of modeling the relationship shown in FIG. 4 with a function (Formula (2)). The figure which shows the example of the parameter for every word calculated | required with the infant word search apparatus 200 from the result of having recorded the word which the infant uttered. The figure which shows the example of the parameter for every word calculated | required with the infant word search apparatus 200 from the result of having recorded the word which the parent judged the infant's understanding. The figure which illustrates the value of "50% reach | attainment age".

  Embodiments of the present invention will be described below with reference to the drawings. The same reference numerals are given to the same components in a plurality of drawings, and the description will not be repeated.

  FIG. 1 shows a functional configuration example of the infant word search device 100 of the present invention. The infant word search device 100 searches for and outputs infant word words that match the question, using the monthly cumulative utterance rate and the user's question as inputs.

[Monthly cumulative utterance rate]
First, the cumulative utterance rate by month will be described. The cumulative utterance percentage monthly, is data consisting of the x acquisition timing vocabulary in a plurality of infants vocabulary learning process and cumulative utterance ratio P _w acquisition word w in. FIG. 2 shows an example where the acquired word w is “Yes”. The horizontal axis acquisition time x, the vertical axis represents the cumulative utterance ratio P _w.

  The data of the cumulative utterance rate is collected for each specific age using, for example, a questionnaire called CDI described in [Non-Patent Document 1]. As another example, for example, it is created based on vocabulary learning data acquired using the web diary method. In the web diary method, when an infant learns a new word (speaks), a specific site on the web is accessed via a network from a mobile phone or a personal computer. (Reference 1 “Tetsuo Kobayashi, Masaaki Nagata (2009)“ Infant Language Development Research Using the Web: Trial of Large-scale Longitudinal Data Collection ”, Proc. 15th Annual Conference of the Language Processing Society, p. 534-537 ", Reference 2" Tetsuo Kobayashi, Masaaki Nagata (March 2010), "Reliability verification of infant vocabulary development data collected on the web", Proc. of the 16th Annual Conference of the Language Processing Society of Japan , P. 403-406). The effectiveness of this method has been scientifically verified.

  The advantage of data acquisition by this method is that it is relatively easy for parents to record, but the difference between the date of recording (the date when the infant learned a new word) and the date of birth of the infant It is possible to calculate the age when an infant learns a new word.

  Note that the vocabulary learning data may be obtained by other methods. For example, all the utterances of an infant may be recorded on an electronic medium such as a digital video recorder, and then analyzed and created. Or you may have a questionnaire answered every predetermined period (for example, every several months), and the result which grasped | ascertained the number of words and the word which the infant uttered newly may be used. Further, the age may be the age of the moon.

  The infant word search device 100 includes a logistic function approximation unit 10, a parameter storage unit 20, and a word search unit 30. The operation flow is shown in FIG. The infant word search device 100 is realized by a predetermined program being read into a computer composed of, for example, a ROM, a RAM, and a CPU, and the CPU executing the program.

The logistic function approximating unit 10 receives the acquisition time x _i of the acquired word w and the cumulative utterance rate P _wi in the vocabulary learning process of a plurality of infants, and inputs the function of the following equation that draws a logistic curve for each acquired word w as the cumulative utterance rate. The parameters c _w and b _w are calculated so that the sum of errors with Pw _i when x = x _i is set is minimized (step S10).

  In modeling the function, a least square method or a likelihood maximization method is used. In this case, since optimization of a non-linear function is necessary, an optimum parameter is obtained using a technique such as the steepest descent method. Both methods are well known and will not be described in detail.

The parameter holding unit 20 stores the parameters c _w and b _w modeled by the logistic function approximating unit 10 (step S20).

The word search unit 30 calculates and searches for a word that matches the input question using the parameters c _w and b _w described above, and outputs it as a desired word (step S30). Here, for example, the question is "What is the word spoken by 80% of 3 year olds?", "What is the word that is difficult to remember (easy)", "What is the word that the infant does not speak", etc. Is assumed.

The word search unit 30 calculates the cumulative utterance rate P _w for all the words w stored in the parameter holding unit 20, for example, selects a word that matches “a word spoken by 80% of a 3-year-old child” as a desired word. Explore as.

FIG. 4 shows the cumulative utterance ratio modeled by the logistic curve calculated by the equation (1). The logistic curve is an S-shaped curve, and shows a change in which the increase amount becomes gentle again after the slope gradually increases after the gentle increase at first. Parameter c _w is the slope of the cumulative incidence percentage of that word (w), b _w represents the offset amount of the logistic curve. When the c _w increases become steeper the slope of the central part of the S-shaped curve, are in a relationship in which the S-shaped curve and b _w increases to move to the larger of the acquisition time x.

Therefore, for example, a word _w having a large parameter c _w is an easy-to-remember word. For example, c <0.005 is given as a specific question. Further, when “a word spoken by 80% of a three-year-old child” is given as a question, the word search unit 30 calculates a cumulative utterance rate P _w for all words at x = 3 years, and P _w ≧ 0.8 Are searched for, and all words that meet the condition are output as desired words. In this way, the infant word search device 100 can easily search for a word that matches the question requested by the user. The logistic curve is so well conform to the relationship between the acquired time x and the true cumulative utterance ratio P _w, the resolution and accuracy of the cumulative utterance ratio P _w can be improved than the conventional method.

The infant word search apparatus 100 may have poor conformity of the logistic curve depending on the word. FIG. 5 shows a logistic curve when the word “barba” is modeled (formula (1)). It does not fit as well as in FIG. This is considered to be due to a phenomenon in which the cumulative utterance rate _Pw reaches a peak around 0.8. Therefore, the infant word search device 200 using a function with better compatibility will be described next.

  FIG. 6 shows a functional configuration example of the infant word search device 200. The operation flow is shown in FIG. The infant word search device 200 has the same functional configuration except that only the function used in the logistic function approximation unit 210 is different from the infant word search device 100.

The logistic function approximating unit 210 receives the acquisition time x _i of the acquired word w and the cumulative utterance rate P _wi in the vocabulary learning process of a plurality of infants and inputs the cumulative utterance rate using the following equation function that draws a logistic curve for each acquired word w. When the parameter _aw is 1 or less, the parameters _aw , _cw, and _bw are calculated as they are. When the parameter _aw is greater than 1, the function of the following equation is modeled by setting a = 1, and the parameter _aw Calculate _w , c _w, and b _w

In the infant word search device 200, _aw may be _aw > 1 due to the addition of the parameter _aw to the function. Since the cumulative utterance rate does not exceed 1, in this case (Yes in step S212), the parameters c _w and b _w are recalculated (step S213) with a _w = 1 (that is, in equation (1)).

The subsequent processing is the same as that of the infant word search device 100. However, since the parameter _aw is added, reference numerals 20 ′ and 30 ′ are indicated for the purpose of distinguishing from the processing of the infant word search device 100.

FIG. 8 shows a logistic curve for the word “barba” by a logistic function in which the parameter _aw is introduced. Than 5, it can be seen that they comply with the cumulative utterance ratio P _w. FIG. 9 shows a logistic curve in which the word “yes” is modeled by the infant word search apparatus 200. It can be seen that the logistic curve (FIG. 4) modeled by the infant word search device 100 is more suitable.

By introducing the parameter _aw , a question such as a> 0.8 can be given to the infant word search device 200. This means "What are the words spoken by more than 80% of infants at around 3 years old (about 1000 days)?"

[Specific example of a parameter _{a w,} c _{w, b} w]
1275 persons with infants and acquisition time x was written for the data of the cumulative utterance ratio P _w of, shown in FIG. 10 the parameters of each word obtained in infant word search device 200. FIG. 10 shows parameters obtained from the result of recording the words uttered by the infant. FIG. 11 shows parameters obtained from a record in which the parent's understanding of the infant is understood. For example, for the word “aa”, the former is a = 0.832787 and the latter is a = 0.986093. This difference comes from the fact that when an infant is about 3 years old, almost all infants can understand "Ah", but only about 83% of children speak this word, ie about 17% This means that the child understands but does not use this word. In this way, by using the two words “a” of the understanding word and the word to be uttered, it is possible to perform a setting such as a word search in which a> 0.95 among the understanding words and a <0.6 that the infant utters. . This setting is understood by most infants, but it is a setting that selects a word that is not used by 40% or more of the childhood.

  When the processing means in the above apparatus is realized by a computer, the processing contents of the functions that each apparatus should have are described by a program. Then, by executing this program on the computer, the processing means in each apparatus is realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only) Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording media, MO (Magneto Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  This program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in a recording device of a server computer and transferring the program from the server computer to another computer via a network.

  Each means may be configured by executing a predetermined program on a computer, or at least a part of these processing contents may be realized by hardware.

  The present invention can be used in the field of customized education for young children.

Claims (5)

As input acquisition time x the cumulative utterance ratio P _w acquisition word w in a plurality of infants vocabulary learning process, to model the function of the following equation to draw a logistic curve for each acquisition word w, calculated parameters c _w and b _w Logistic function approximation part

A parameter storage unit for storing the parameters c _w and b _w ;
Acquisition time x, the cumulative utterance ratio P _w, parameter c _w, the word that matches one or more directed questions b _w, and search is calculated using the parameters c _w and b _w described above, the desired word A word search unit to output as
An infant word search device comprising: As input acquisition time x the cumulative utterance ratio P _w acquisition word w in a plurality of infants vocabulary learning process, and the following equation models the function of drawing a logistic curve for each of the acquisition word w, the parameter a _w is 1 or less In this case, the parameters _aw , _cw, and _bw are calculated as they are, and when the parameter _aw is larger than 1, the function of the following equation is modeled by setting _aw = 1, and the parameters _aw , _cw, and _bw are calculated. Logistic function approximator,

A parameter storage unit for storing the parameter _{a w} and _{c w} and _{b w,}
Acquisition time x, the cumulative utterance ratio _{P w,} parameter _{a w,} c w, the word that matches one or more directed questions _{b w,} calculated using the parameters _{a w} and _{c w} and _{b w} described above A word search unit for searching and outputting as a desired word;
An infant word search device comprising: As input acquisition time x the cumulative utterance ratio P _w acquisition word w in a plurality of infants vocabulary learning process, to model the function of the following equation to draw a logistic curve for each acquisition word w, calculated parameters c _w and b _w Logistic function approximation process to

A parameter storage process for storing the parameters c _w and b _w in the parameter storage unit;
Acquisition time x, the cumulative utterance ratio P _w, parameter c _w, the word that matches one or more directed questions b _w, calculated using the parameters c _w and b _w stored in the parameter storage unit A word search process for searching and outputting as a desired word;
An infant word search method executed by a computer . As input acquisition time x the cumulative utterance ratio P _w acquisition word w in a plurality of infants vocabulary learning process, and the following equation models the function of drawing a logistic curve for each of the acquisition word w, the parameter a _w is 1 or less In this case, the parameters _aw , _cw, and _bw are calculated as they are, and when the parameter _aw is larger than 1, the function of the following equation is modeled by setting _aw = 1, and the parameters _aw , _cw, and _bw are calculated. Logistic function approximation process,

The parameter storage unit, and a parameter storage step of storing the parameter _{a w} and _{c w} and _{b w,}
Acquisition time x, the cumulative utterance ratio _{P w,} parameter _{a w,} c w, the word that matches one or more directed questions _{b w,} parameters stored in the parameter storage unit _{a w} and _{c w} and _{b w} A word search process that calculates and searches using, and outputs as a desired word;
An infant word search method executed by a computer .   A program for causing a computer to function as the infant word search device according to claim 1.

Images