JP7488702B2 - Information Management System - Google Patents

Description

The present invention relates to an information management system.

There are high expectations for highly efficient materials development through materials informatics (MI) and process informatics (PI). Handling large amounts of data, including data from experiments conducted by others, is effective because it increases the likelihood of finding new correlations between parameters and optimal parameter values with a small number of experiments. Meanwhile, materials development typically involves many work processes to achieve the desired performance or functionality, making MI that takes into account the time-series information between these parameters.

Patent Document 1 also discloses a method for extracting important keywords from the work screen when a user is developing a product, identifying the user's development field from the extracted important keywords, and generating information to propose to the user.

JP 2018-67230 A

However, the technology disclosed in Patent Document 1 does not relate to the field of materials development, and when experiments are conducted through a variety of processes, it is difficult to provide users with information that takes into account the time series.

The objective of the present invention is to provide an information management system in the field of materials development that enables efficient process design using past experimental data.

To achieve the above object, the present invention comprises a storage unit that stores experimental data of materials, a time series extraction unit that extracts time series information of a material processing process from the experimental data, a keyword extraction unit that extracts candidate keywords related to the material processing process from the experimental data, and an index creation unit that creates an index based on the time series information and the candidate keywords.

The present invention provides an information management system that enables efficient process design using past experimental data in the field of materials development.

A diagram of the information management system. FIG. 2 is a diagram showing the configuration of a material-related database group. FIG. 4 is a diagram showing details of a process time series extraction unit. FIG. 13 is a diagram showing an example of a table output by a table output unit. FIG. 13 is a diagram showing an example of time-series information of a material processing process. FIG. 11 is a flow diagram showing steps in which a process time series extraction unit extracts time series information based on a table included in the experimental data and a group of material-related DBs. FIG. 4 is a diagram showing details of a keyword extraction unit. FIG. 13 is a diagram showing an example of keywords related to material treatment processes. FIG. 13 is a diagram showing details of a sample identifier extraction unit. FIG. 13 shows an example of a sample identifier for an experimental subject. FIG. 4 is a diagram showing an example of an index created by an index creating unit.

An embodiment of the present invention will be described below. FIG. 1 is a configuration diagram of an information management system according to this embodiment. The information management system according to this embodiment supports research and development related to materials, and as shown in FIG. 1, includes an experimental data storage unit 100, a process time series extraction unit 200, a keyword extraction unit 300, a sample identifier extraction unit 400, a data formatting unit 500, an index creation unit 600, and a data catalog unit 700.

The experimental data storage unit 100 stores experimental data on materials, and related files for each project are uploaded by experimenters, etc. The files stored include documents and tables of experimental plans, etc.

The process time series extraction unit 200 extracts time series information of material processing processes from the experimental data. It extracts process names and the like from documents and tables contained in files in the experimental data storage unit 100, estimates the order of the processes based on their relative positions, and displays and registers the results of the estimation. Details of the process time series extraction unit 200 will be described later using Figures 3 to 6.

The keyword extraction unit 300 extracts candidate keywords related to material processing processes from the experimental data, and estimates keywords from documents and tables contained in files in the experimental data storage unit 100, such as frequently used terms, terms found in titles, and terms set by the experimenter, and displays and registers the results of the estimation. Details of the keyword extraction unit 300 will be described later using Figures 7 and 8.

The sample identifier extraction unit 400 extracts material identifiers from the experimental data, and extracts numbers such as #1, #2 (No. 1, No. 2 or sample a, sample b, etc.) that identify the process through which the sample was made from documents and tables contained in files in the experimental data storage unit 100. Details of the sample identifier extraction unit 400 will be described later using Figures 9 and 10.

The data reforming unit 500 converts the format of documents and tables contained in files in the experimental data storage unit 100 into standardized data and stores it. The index creation unit 600 creates indexes based on time-series information, keyword candidates, and identifiers. The data catalog unit 700 organizes and stores the information obtained by the data reforming unit 500 and index creation unit 600 as a data catalog so that users of the information management system can search for it. Note that by creating an index using keywords as well as time-series information, the accuracy of material searches by users is improved, but identifiers are not essential for creating an index.

The information management system of this embodiment further includes a data search unit 701, a parameter relationship analysis unit 702, a unit conversion unit 703, a data integration unit 704, and a database storage unit 705. The data search unit 701 searches for information stored in the data catalog unit 700 based on keywords related to materials entered by the user, and outputs an outline of the experimental results in order of the most relevant. The parameter relationship analysis unit 702 analyzes the relationship between the parameters of each process. For example, if the heat treatment process is performed after the mixing process, it analyzes the relationship between the heat treatment temperature and the mixing time. The unit conversion unit 703 converts the data into a unit system that is easy to use according to the user's preferences. The data integration unit 704 integrates the above-mentioned data, and the database storage unit 705 stores the integrated data as a database of MI and PI. In the information management system of this embodiment, the configuration from the data search unit 701 to the database storage unit 705 automatically stores the data in the database, but the user may do so manually.

FIG. 2 is a diagram showing the configuration of the material-related database group 800. As shown in FIG. 2, the material-related database group 800 of this embodiment includes a process name database 801, a parameter name database 802, a unit symbol database 803, and an apparatus/equipment database 804. The process name database 801 stores specific names of processes used in the material field, such as heat treatment, mixing, separation, and reaction. The parameter name database 802 stores names of parameters, such as flow rate, time, and temperature, that are necessary when defining the process. The unit symbol database 803 stores units, such as ml and °C, related to the parameter. The unit symbol database 803 also stores symbols commonly used to express the parameter, such as _Ta for temperature. The apparatus/equipment database 804 stores information about devices and equipment used in material experiments.

The material-related database group 800 of this embodiment also functions as an integrated database that organizes the correlations between the information stored in each database. For example, certain processes are often related to specific parameters, units, and symbols, and information including such correlations is stored in the material-related database group 800. In addition, a specific device may be used to perform a certain process, and in such cases, information is stored in the material-related database group 800 with the process name and the device linked. Note that the material-related database group 800 may be stored in an information management system or an external server that can be connected via a network, as long as it is registered in advance.

Figure 3 is a diagram showing details of the process time series extraction unit 200. The process time series extraction unit 200 of this embodiment reads out tables from the experimental data stored in the experimental data storage unit 100 to obtain the relative positions of the process names, etc. contained in the tables, and also reads out documents to obtain the relative positions of the process names, etc. contained in the documents. The process time series extraction unit 200 of this embodiment also uses these obtained relative positions to estimate time series information of the material processing process, and displays the results to the experimenter, etc. After the experimenter, etc. has confirmed the data and registered it, the process time series extraction unit 200 stores the estimated time series information.

The specific configuration of the process time series extraction unit 200 will now be described with reference to FIG. 3. The process time series extraction unit 200 of this embodiment first includes a table extraction unit 201, a table conversion unit 202, a table output unit 203, a confirmed table storage unit 204, a table header acquisition unit 205, and a header layout acquisition unit 206, and acquires information required to estimate the process time series from tables included in the experimental data, particularly from the contents of their headers. This takes into account that material processing processes are generally written from left to right in the table header, starting from the upstream process.

The table extraction unit 201 extracts data corresponding to a table from the experimental data stored in the experimental data storage unit 100. Here, since the data format of the table may differ depending on the experimenter, the table conversion unit 202 converts it into a standard table. The table output unit 203 outputs the table converted into a standard format to the experimenter's terminal. Note that the experimenters in this embodiment include not only those who actually performed the experiment, but also those who did not perform the experiment but created tables or experimental plans, people in a position to supervise these people, and people who work under the supervision of these people.

Figure 4 is a diagram showing an example of a table output by the table output unit 203. As shown in Figure 4, the header of the table lists the process names, such as A reaction, B mixing, and C evaluation, and for each sample identifier, the parameter amounts corresponding to each process name are listed. If the experimenter checks the output table and there are no problems, it is registered in the confirmed table storage unit 204 as confirmed data.

Next, the table header acquisition unit 205 acquires the contents of the header from the confirmed table. Then, the header placement acquisition unit 206 searches for information such as the process name in the header, and acquires the placement of the process name extracted by the search. At this time, the header placement acquisition unit 206 can also refer to the material-related database group 800 when searching for the process name. For example, in the case of the table shown in FIG. 4, the header placement acquisition unit 206 acquires the placement of the process name such as A reaction first from the left, then B mixing, then C evaluation, and the placement of the parameter name such as flow rate first from the left, then time, then yield. In addition to the process name and parameter, the header placement acquisition unit 206 can also identify the physical quantity, unit, or symbol, or the character string such as the experimental apparatus and experimental method, using the material-related database, and acquire the positional relationship in the header. This is because if there is information such as the physical quantity and unit, it is possible to know to some extent which process it is related to by referring to the material-related database.

As shown in FIG. 3, the time series extraction unit of this embodiment also includes an experimental plan layout acquisition unit 207, which acquires information required for estimating the time series of a process from the contents of the experimental plan document. In general, it is considered that a material processing process is described in a document in order from the upstream process. The experimental plan layout acquisition unit 207 of this embodiment also refers to the material related database group 800 to search for information such as process names in the document and acquire the layout of the process names extracted by the search. The experimental plan layout acquisition unit 207 can also acquire positional relationships in the document using parameter names, physical quantities, units, or symbols, or experimental equipment, experimental methods, etc., in the same way as the header layout acquisition unit 206. Note that when the experimental equipment and the experimental method correspond to each other, the experimental plan may only describe either the experimental equipment or the experimental method.

The process time series estimation unit 208 estimates the time series of the material processing process based on the positional relationship of the process name, etc. in the header acquired by the header layout acquisition unit 206 and the positional relationship of the process name, etc. in the experiment plan acquired by the experiment plan layout acquisition unit 207. The estimation result of the process time series estimation unit 208 is output to the experimenter's terminal by the process time series display unit 209.

Figure 5 is a diagram showing an example of time series information of a material processing process displayed by the process time series display unit 209. As shown in Figure 5, the order of processes is displayed, such as first A weighing, then B reaction, then C fractional distillation, and finally D evaluation. Here, the evaluation process is performed by various analyses using X-ray devices and scanning electron microscopes, including the texture and structure of the material, so there are usually multiple types of evaluation. Therefore, Figure 5 shows that in addition to D evaluation, E evaluation was also performed for the same sample identifier. Note that, other than evaluation, multiple processes may exist at the same time, such as B reaction and F reaction being performed together. In other words, the time series information may be displayed in a tree shape including parallel processes, as long as the order of the processes can be specified.

The estimated process time series is checked by the experimenter, and if there are no problems, it is registered in the process time series storage unit 210. If the displayed process time series is incorrect, the experimenter corrects the data. The history of corrections made by the experimenter is registered in the correction method storage unit 211 and is used for subsequent estimations via the learning unit 212.

Having such a process time series extraction unit 200, users of the information management system can grasp the context between processes, making it easy to infer the governing factors for realizing the desired performance and functionality when utilizing data, enabling efficient process design.

Figure 6 is a flow diagram showing the steps in which the process time series extraction unit 200 extracts time series information based on tables included in the experimental data and the material-related DB group. First, the table extraction unit 201 extracts tables included in the experimental data stored in the experimental data storage unit 100 (step S901). Next, the table conversion unit 202 formats the extracted tables into standard data, and the table output unit 203 displays the formatted tables on the experimenter's terminal (step S902). The experimenter then checks that the output table is error-free (step S903), and if there are no errors, it is stored as a finalized table (step S904). If there are errors, the experimenter corrects the output table (step S905), and the corrected table is stored as a finalized table.

Next, the table header acquisition unit 205 acquires header data from the confirmed table (step S906). Then, the header internal arrangement acquisition unit 206 refers to the material related database group 800 and acquires the relative position of the data (step S907). Furthermore, the process time series estimation unit 208 estimates the time series of the material processing process based on the relative position (step S908). After that, the process time series display unit 209 displays the estimated time series on the experimenter's terminal (step S909). Then, the experimenter checks that there are no errors in the displayed time series (step S910), and if there are no errors, it is stored in the process time series storage unit 210 (step S911). If there is an error, the experimenter corrects the process time series (step S912), and the corrected one is stored in the process time series storage unit 210.

Figure 7 is a diagram showing details of the keyword extraction unit 300. The keyword extraction unit 300 of this embodiment estimates keyword candidates from documents, tables, etc. included in the experimental data stored in the experimental data storage unit 100, and displays the results to the experimenter. When the data is registered as a result of confirmation by the experimenter, the keyword extraction unit 300 stores the estimated keywords.

The keyword extraction unit 300 of this embodiment includes a user-defined information acquisition unit 301, a frequently-used term acquisition unit 302, a table header acquisition unit 303, a physical quantity/unit symbol acquisition unit 304, an apparatus/method acquisition unit 305, and a substance information (organization/compound/structural formula, etc.) acquisition unit 306, which estimate keyword candidates. The user-defined information acquisition unit 301 acquires information defined when an experimenter customizes and defines a keyword. For example, it is used when an experimenter wants to record his/her own name or the date of an experiment. The frequently-used term acquisition unit 302 acquires terms that are written a certain number of times or more in documents included in the experimental data as frequently-used terms. The table header acquisition unit 303 acquires terms that are written in the titles of rows and columns, such as headers, of tables included in the experimental data as being highly likely to be keywords. The physical quantity/unit symbol acquisition unit 304 acquires the physical quantity (e.g., heat quantity) or the unit of the physical quantity (e.g., calorie) or commonly used symbol (e.g., Q) contained in the header of the document or table of the experimental data. The equipment and method acquisition unit 305 acquires the equipment and method used in the experiment, and the material information acquisition unit 306 acquires the composition, compound, structural formula, etc. of the material. Note that when the physical quantity/unit symbol acquisition unit 304 and the equipment and method acquisition unit 305 acquire information related to the physical quantity or the equipment, the material related database group 800 is also referenced.

Based on information acquired by the frequently used term acquisition unit 302 and the like, the keyword estimation unit 307 estimates at least one of the substance names (including abbreviations and common names), compositions, chemical formulas, and structural formulas described in the experimental data as keyword candidates. At this time, the keyword estimation unit 307 may refer to the technical term database 810, the synonym database 820, the ontology database 830, and the like. The estimation results of the keyword estimation unit 307 are then displayed on the experimenter's terminal by the keyword display unit 308.

Figure 8 shows an example of keywords related to material processing processes displayed by the keyword display unit 308. The displayed keywords related to material processing processes are checked by the experimenter and, if there are no problems, are registered in the keyword storage unit 309. Furthermore, if the displayed terms include ones that do not correspond to keywords, the experimenter corrects the data. The history of corrections made by the experimenter is registered in the correction method storage unit 310 and is used for subsequent estimations via the learning unit 311.

By having such a keyword extraction unit 300, even if a user of the information management system enters, for example, "Device A, Compound B" to search, the device and compound are linked to the experimental data as keywords, so experimental data containing these keywords can be output.

Figure 9 is a diagram showing the details of the sample identifier extraction unit 400. First, the sample identifier estimation unit 401 estimates the sample identifier contained in documents, tables, etc. contained in the experimental data stored in the experimental data storage unit 100 while referring to the material-related database group 800. The estimation result of the sample identifier estimation unit 401 is then displayed on the experimenter's terminal by the sample identifier display unit 402.

Figure 10 shows an example of the sample identifier of the experimental subject displayed by the sample identifier display unit 402. The displayed sample identifier is checked by the experimenter and, if there is no problem, is registered in the sample identifier storage unit 403. If there is an error in the displayed sample identifier, the experimenter corrects the data. The history of corrections made by the experimenter is registered in the correction method storage unit 404 and is used for subsequent estimations via the learning unit 405.

Figure 11 is a diagram showing an example of an index created by the index creation unit 600. The index creation unit 600 of this embodiment creates an index based on the time series information extracted by the process time series extraction unit 200, the keyword candidates extracted by the keyword extraction unit 300, and the sample identifier extracted by the sample identifier extraction unit 400. As shown in Figure 11, for example, a sample with an identifier #1 is created in the order of process A, process B, ..., and the parameter names of each process are temperature, flow rate, ..., and the corresponding table data is stored in sheet Y of file X, and the keywords are reaction A and compound B. In addition, even if a user of the information management system wants to search for an experiment performed within a certain period of time, for example, it is possible to identify the corresponding experimental data if date information is registered as a keyword. Note that the index creation unit 600 in this embodiment creates an index that also includes a sample identifier, but an index may be created using only process time series information and keywords.

The above-mentioned embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to having all of the configurations described.

100 Experimental data storage unit 200 Process time series extraction unit 201 Table extraction unit 202 Table conversion unit 203 Table output unit 204 Confirmation table storage unit 205 Table header acquisition unit 206 Header layout acquisition unit 207 Experimental plan layout acquisition unit 208 Process time series estimation unit 209 Process time series display unit 210 Process time series storage unit 211 Correction method storage unit 212 Learning unit 300 Keyword extraction unit 301 User-defined information storage unit 302 Frequently used term acquisition unit 303 Table header acquisition unit 304 Physical quantity/unit symbol acquisition unit 305 Equipment/method acquisition unit 306 Material information (composition/compound/structural formula) acquisition unit 307 Keyword estimation unit 308 Keyword display unit 309 Keyword storage unit 310 Correction method storage unit 311 Learning unit 400 Sample identifier extraction unit 401 Sample identifier estimation unit 402 Sample identifier display unit 403 Sample identifier storage unit 404 Correction method storage unit 405 Learning unit 500 Data shaping unit 600 Index creation unit 700 Data catalog unit 701 Data search unit 702 Inter-parameter relationship analysis unit 703 Unit conversion unit 704 Data integration unit 705 Database storage unit 800 Material-related database group 801 Process name database 802 Parameter name database 803 Unit symbol database 804 Equipment database

Claims (9)

a storage unit for storing experimental data of materials;
a time series extraction unit for extracting time series information of a material processing process from the experimental data;
a keyword extraction unit that extracts candidates for keywords related to the material treatment process from the experimental data;
and an index creation unit that creates an index for searching for desired experimental data by associating the experimental data with the time-series information or the keyword candidates. 2. The information management system according to claim 1, wherein the time series extraction unit acquires relative positions of a plurality of process names included in the experimental data while referring to a preregistered material related database, and extracts an order of the processes as the time series information based on the acquired relative positions. 3. The information management system according to claim 2, wherein said time series extraction unit has a display unit that displays said time series information and enables a user to register whether or not said time series information needs to be corrected. 2. The information management system according to claim 1, further comprising a sample identifier extraction unit that extracts an identifier of a target sample from the experimental data, wherein the index creation unit creates the index based on the identifier. The information management system of claim 1, wherein the time series extraction unit searches for character strings of the device and/or experimental method described in the experimental data, and extracts the time series information based on the relative positions of the character strings. The information management system of claim 1, wherein the keyword extraction unit estimates the titles of rows or columns in a table described in the experimental data as candidates for the keyword. The information management system of claim 1, wherein the keyword extraction unit estimates at least one of the substance name, composition, chemical formula, and structural formula described in the experimental data as a candidate for the keyword. 2. The information management system according to claim 1, wherein the keyword extraction section displays the estimated keyword candidates, and if an error is found, the keyword candidates are corrected by an experimenter . 9. The information management system according to claim 8 , wherein said keyword extracting section has a learning section that learns based on a history corrected by said experimenter .

Images