WO2021090389A1

WO2021090389A1 - Building information processing device

Info

Publication number: WO2021090389A1
Application number: PCT/JP2019/043414
Authority: WO
Inventors: 嘉人遠藤; 圭一庄司
Original assignee: 三菱電機ビルテクノサービス株式会社
Priority date: 2019-11-06
Filing date: 2019-11-06
Publication date: 2021-05-14
Also published as: CN114730443A; CN114730443B; JP7160503B2; JPWO2021090389A1

Abstract

A building information processing device (70) is provided with a machine learning processing unit (86), an extraction unit (82), and AI (84). The machine learning processing unit (86) carries out machine learning with respect to the AI (84) using a characteristic amount dictionary (74), which includes expressions regarding a defect in equipment or a component in a building. Thus, the AI (84) carries out learning in which a characteristic expression that is set in a characteristic expression dictionary (78) and that expresses a maintenance control characteristic regarding equipment or a component is selected. An extraction unit (82) extracts, for each of a plurality of text strings describing a communication or a report related to a defect in equipment or a component in the building, an expression regarding the defect in the equipment or component. The AI (84) then is used to select a text string, and a characteristic expression corresponding to the extracted expression regarding the defect in the equipment or component.

Description

Building information processing equipment

The present invention relates to a building information processing device that handles building operations or defects in building equipment or parts.

In a building, in addition to the operation work of the building, maintenance management work including dealing with defects is usually carried out for various equipment or parts such as elevators, air conditioners, and electrical equipment. In the event of a defect, or when a defect is addressed, a document is generally created to contact or report.

Patent Document 1 describes an information processing device that assists an operation for contacting a monitoring center or the like when a user such as an elevator owner, administrator, or passenger encounters a problem. In the information processing device, the user can select the defect state from the items registered in advance, and can also freely describe it.

Patent Document 2 describes a report creation system when a worker who performs maintenance and inspection of an elevator works on a work target item. For each work item, the worker sequentially inputs work item specific information, work results, and comments by voice. The management server creates a report by documenting the voice, extracting keywords, and classifying them according to work target items.

If you are notified of a problem, the worker will be contacted.

Patent Document 3 documents a voice call sent to a monitoring center from a user such as an elevator or escalator administrator or a passenger, and processes the information to send a message to a worker who performs maintenance and inspection. It is described to create. In information processing, when a preset keyword is extracted from a document, a template message corresponding to the keyword is prepared. In addition to inputting keywords extracted from the document in the blank part provided in the template message, the transmitted message is completed when the operator inputs it.

Responses to defects are also being made in fields other than buildings.

Patent Document 4 describes a system that makes it possible to access similar documents in the past by hierarchically classifying documents describing the occurrence of computer malfunctions based on the similarity.

JP-A-2019-1618 Japanese Unexamined Patent Publication No. 2015-75792 JP-A-2018-156134 Japanese Unexamined Patent Publication No. 2005-267351

As described in

Patent Documents

1 and 2, when a defect occurs in the equipment or parts of a building, the building owner, tenant, visitor, or other user, or the building maintenance manager, contacts or The report will be made. These contact or report documents (or strings) contain useful information for building maintenance. Therefore, it would be convenient if information according to the defect could be extracted from such a contact or report document or character string. However, in general, since the equipment or parts of a building are diverse, it is complicated to extract information according to a defect, which requires time and cost.

Since Patent Document 3 documents audio, it handles documents or character strings that are close to free description. However, Patent Document 3 covers maintenance management work of only elevators or only escalator, and since the content of maintenance management work is limited, it is relatively easy to select a template message based on keywords. Conceivable. Moreover, in Patent Document 3, since the worker directly confirms and creates a message, a correct message is finally created even if the accuracy of selecting the template message is poor. Therefore, it is difficult to apply the technique of Patent Document 3 to various facilities or parts of a building.

Patent Document 4 only deals with the limited content of computer malfunctions, and also only classifies documents or character strings having a fixed format. Therefore, it is difficult to apply the technique of Patent Document 4 to defects in building equipment or parts.

An object of the present invention is to obtain a phrase indicating a characteristic in maintenance management from a character string in which a contact or report relating to a building operation or a defect in a building's equipment or parts is described.

In the building information processing apparatus according to the present invention, characteristic terms representing the characteristics of maintenance management of the building, equipment or parts are selected by machine learning using the terms related to the operation of the building or the defects of the equipment or parts of the building. Extract words and phrases about the operation of the building or the malfunction of the equipment or parts from the character string in which the AI learned to do and the contact or report regarding the operation of the building or the malfunction of the equipment or parts of the building are described. The extraction means is provided, and the selection means for selecting the characteristic phrase corresponding to the extracted phrase regarding the operation of the building or the defect of the equipment or part is provided.

In one aspect of the invention, from multiple documents in different formats created by the user or maintenance manager of the building and describing communications or reports relating to the operation of the building or the malfunction of equipment or parts of the building. , The preprocessing means for acquiring a plurality of the character strings is further provided, the extraction means extracts words and phrases about a defect of the equipment or a component for each character string, and the selection means uses the AI. , The characteristic phrase is selected for each of the character strings.

In one aspect of the present invention, a setting means for setting at least a part of words and phrases regarding the operation of the building or the malfunction of the equipment or parts is further provided based on the instruction of the manager of the building information processing apparatus.

In one aspect of the present invention, the phrase regarding the operation of the building or the defect of the equipment or part describes the name of the building, equipment or part, and the operation or defect occurring in the building, equipment or part. Descriptive words and phrases are included.

In one aspect of the present invention, the descriptive phrase is a phrase that describes operation or defect identification, operation or defect response request, or operation or defect response implementation.

In one aspect of the present invention, the characteristic phrase is a phrase that summarizes the operation or malfunction of the building, equipment, or part.

In one aspect of the present invention, the characteristic phrase is a phrase that predicts the operation or defect identification of the building, equipment or part, the request for dealing with the operation or the defect, or the implementation of the countermeasure for the operation or the defect.

In one aspect of the present invention, when the character string or the document from which the character string is acquired contains a predetermined word / phrase, the word / phrase or a word / phrase corresponding to the word / phrase is referred to as the character string. It is provided with an output means for outputting together with the above-mentioned characteristic phrase.

According to the present invention, it is possible to select a characteristic phrase that expresses the characteristics of maintenance management of the building, equipment, or parts from a character string that describes the operation of the building or the defect of the equipment or parts of the building. Become. This makes it possible to create a document using, for example, characteristic words. Further, for example, it is possible to classify a character string for a defect according to a characteristic phrase.

It is a schematic block diagram of the building system which concerns on embodiment. It is a figure which shows the schematic structure of a building. It is a figure which shows the schematic structure of the building information processing apparatus. It is a figure which shows the example of the daily report which is a report document which described a defect. It is a figure which shows the example of the communication document which described the defect. This is an example of an inquiry / response character string created from a document. It is a figure which shows the relationship between a feature quantity and a characteristic phrase. This is an example of selecting characteristic words based on the features extracted from the inquiry / response character string. This is an example of a rule-based phrase. It is a figure which added the rule-based phrase extracted from the document to FIG.

FIG. 1 is a diagram showing a schematic configuration of a building system 10 according to an embodiment. In the figure, three

buildings

20a, 20b, and 20c (hereinafter, simply referred to as building 20 without distinguishing each other) are shown as examples of a large number of buildings. The building 20 is maintained and managed by the building maintenance company, including operation and troubleshooting. The building 20 is communicably connected to the building information processing device 70 through the communication network 60. A PC 100 that performs data processing is connected to the building information processing device 70. The building information processing device 70 and the PC 100 are installed in, for example, a building maintenance company.

FIG. 2 is a diagram showing in detail the building 20 shown in FIG. The building 20 is equipped with an air conditioner 22, a water supply / drainage facility 24, an electric facility 26, an elevator 28, and the like.

In addition, a central monitoring room 30 is provided on the basement of the building 20. The central monitoring room 30 is provided with a central monitoring board 32 and a PC 34. The central monitoring panel 32 is a computer that is communicably connected to the air conditioning equipment 22, the water supply / drainage equipment 24, the electrical equipment 26, and the like to control and monitor the operation.

The PC 34 is connected to the remote monitoring device of the elevator 28 to control and monitor the elevator 28, and also analyzes the operation data of the elevator. The PC 34 is also connected to the central monitoring board 32 and analyzes the data output by the central monitoring board 32. Further, the PC 34 is used by

maintenance personnel

36 and 38 who perform maintenance management work based on the central monitoring room 30 to create a business report such as a daily report. Further, the PC 34 is also used to receive from the user of the building 20 a communication document regarding the operation of the building or a communication document regarding a defect related to equipment and parts by e-mail or the like.

In FIG. 2, it is assumed that a defect has occurred in the part 24a of the water supply / drainage facility 24. Then, the tenant 40, who is the user of the building 20, finds a defect in the component 24a and contacts the central monitoring room 30 through the mobile phone 42. In the central monitoring room 30, the maintenance staff 36 is contacted by the tenant 40 through the telephone 44. As a result, the maintenance worker 36 will contact himself or another maintenance worker 38 to repair the parts 24a and the like.

Further, in FIG. 2, the maintenance staff 38 confirms a defect in the electrical equipment 26 that has been separately contacted. Based on the confirmation result, the maintenance staff 38 carries out necessary repairs and the like.

FIG. 3 is a diagram showing a detailed configuration of the building information processing device 70 shown in FIG. The building information processing device 70 is constructed by controlling computer hardware including a memory and a processor with software including an operating system and an application program. The building information processing apparatus 70 may be constructed by using a single computer hardware, or may be constructed by using a plurality of computer hardware connected to the network.

Depending on the application program, the building information processing device 70 includes a feature amount dictionary setting unit 72, a feature amount dictionary 74, a characteristic word setting unit 76, a characteristic word dictionary 78, a document input unit 80, a preprocessing unit 81, an extraction unit 82, and AI 84. , Machine learning processing unit 86, characteristic phrase acquisition unit 88, output unit 90, rule-based dictionary setting unit 92, rule-based dictionary 94, and rule-based phrase acquisition unit 96 are constructed.

The feature amount dictionary setting unit 72 is an example of the setting means, and the administrator of the building information processing apparatus 70 inputs and sets the feature amount dictionary 74. The feature quantity dictionary 74 is a dictionary in which a plurality of words and phrases are stored for defects in equipment or parts of a building. A defect is a state in which an abnormality, failure, trouble or deterioration of equipment or parts has occurred or is presumed to occur. Defects include situations where there is no problem with the equipment or the parts themselves, but in operation, problems will occur unless the settings are changed.

The words and phrases about defects include the names of equipment and parts of the building and descriptive words and phrases that describe the defects that occur in these equipment and parts. This is because it becomes easier to understand the defect by including the target in which the defect has occurred and a phrase that describes what the defect is.

The names of equipment and parts include general names, manufacturer product numbers, and other names that describe the equipment and parts to the extent that

maintenance personnel

36 and 38 can identify them. The descriptive phrase that describes the defect includes a phrase that describes at least one of the identification of the defect, the request for dealing with the defect, and the implementation of the countermeasure for the defect. The phrase that identifies the defect is a phrase that describes the state or cause of the defect, such as an abnormal noise or a broken light bulb. The phrase for requesting the handling of a defect is a description indicating a request for confirmation, repair, replacement, etc. of the defect, such as requesting the temperature to be adjusted or repaired. The phrase indicating that the defect has been dealt with is a description indicating that the defect has been dealt with, such as confirmation, repair, or replacement. In addition to one word (regardless of part of speech such as noun, verb, adjective, etc.), the phrase includes a phrase in which two or more words are connected. The phrase may include a concise sentence.

The feature amount dictionary 74 can include other characteristic words and phrases regarding defects in building equipment or parts. Specifically, the phrase indicating the time related to the defect such as the date and time of inquiry of the defect, the date and time of dealing with the defect, or the details of the defect such as the floor where the defect occurred, the room where the defect occurred, the name of the equipment where the defect occurred, and the part name. The phrase representing may be adopted. As another example, there are words and phrases that represent the person who dealt with the problem, such as the name of the company that dealt with the problem and the name of the person.

The words and phrases set and registered in the feature dictionary 74 are greatly related to the accuracy of the words and phrases selected by AI84. Therefore, a part of the feature amount dictionary 74 can be created by a process of automatically selecting the feature amount, but even in that case, the administrator can use the feature amount dictionary setting unit 72 to create the validity of the feature amount. Is examined, and the feature amount dictionary 74 is set. In particular, if a person familiar with building maintenance management sets the feature dictionary 74, it can be expected that the selection accuracy by AI84 will be improved.

In general, since there are synonyms or synonyms having similar meanings in words and phrases, synonyms or synonyms are also set in the feature dictionary 74.

In the characteristic phrase setting unit 76, the administrator of the building information processing device 70 inputs and sets the characteristic phrase dictionary 78. A part of the characteristic phrase dictionary 78 may be automatically created by using, for example, the feature amount dictionary 74. However, even in that case, the administrator examines the validity of the characteristic phrase through the characteristic phrase setting unit 76 and sets the characteristic phrase dictionary 78.

The characteristic phrase dictionary 78 stores a plurality of characteristic phrases that represent the characteristics of maintenance management of equipment or parts. The characteristic phrase may be a phrase that summarizes a defect in equipment or parts. For example, a concise combination of the name of a facility or part and its condition summarizes the defect. In addition, the characteristic phrase may include a phrase that describes identification of a defect in equipment or parts, a request for countermeasures for the defect, or implementation of countermeasures for the defect. The words and phrases set in the characteristic word dictionary 78 are associated with the words and phrases set in the feature amount dictionary 74. This association relationship is used in machine learning, which will be described later.

In general, since there are synonyms or synonyms having similar meanings in words and phrases, it is possible to set and register synonyms or synonyms in the characteristic phrase dictionary 78. However, as will be described later, the characteristic phrase dictionary 78 defines words and phrases selected by AI 84. In order to prevent the words and phrases selected by AI 84 from being ambiguous by synonyms or synonyms, it is conceivable that the characteristic word dictionary 78 does not have synonyms or synonyms.

The document input unit 80 acquires a document in which a contact or report regarding a defect of equipment or parts is described from the building 20 shown in FIG. 1 or a storage server. A communication document or report document about a defect is a document created by a building user (meaning a person who uses the building such as an owner, a tenant, a visitor, etc.) or a maintenance manager such as a

building maintenance staff

36 or 38. , A device that informs you that there is a problem with equipment or parts. In general, report documents are more likely to be prepared under some responsibility or authority than correspondence documents. However, since communication documents and report documents have similar characteristics, they may not always be clearly distinguishable.

The communication document or report document may be created by the user and sent to the maintenance manager of the building, for example. Further, the communication document or the report document may be created by, for example,

maintenance personnel

36, 38 and transmitted to a building user or another maintenance manager. The correspondence or report document is created for various purposes such as notifying that a defect has occurred, recording the occurrence of a defect, requesting that the defect be addressed, and notifying that the defect has been addressed. It is possible that it will be done.

Communication documents and report documents are typically created electronically using word processing software. However, the communication document and the report document may be electronic documents in which voice is converted into characters by voice recognition processing, or handwritten characters may be converted into characters by OCR (Optical Character Recognition).

The preprocessing unit 81 is an example of the preprocessing means, and extracts a character string in which a defect of equipment or a component is described from each document acquired by the document input unit 80 and summarizes it as a list. Generally, the correspondence or report document input from the document input unit 80 is described in various formats. Therefore, when a plurality of defects are listed in one document, the preprocessing unit 81 performs a process of selecting a portion in which each defect is described. The selection can be made automatically or according to preset settings, recognizing the formatting features of the document (eg, line boundaries). It is also possible to make a selection by interpreting the meaning of each area using, for example, natural language processing.

Next, when the description of the defect is described in a plurality of columns, the preprocessing unit 81 performs a process of extracting a character string in which the descriptions in those columns are integrated. For example, in many correspondence documents or report documents, an item corresponding to "inquiry content" and an item corresponding to "handling an inquiry" are prepared. Therefore, a character string that integrates the description character strings of these two items (or items corresponding to their synonyms and synonyms) is extracted.

In this embodiment, the term character string refers to a series of characters, numbers, symbols, spaces, and the like. The term character string is a concept similar to the term sentence, but it does not necessarily have to be composed of sentences. For example, a character string includes a series of words and phrases that are not in the form of a sentence because a plurality of entry fields are combined. The term document (sometimes called text) is used to describe a relatively large unit, such as an electronic file. On the other hand, in the present embodiment, the term character string is used with the intention that it may be a part of the elements extracted from the document. For example, when an electronic file is formed for each character string, the character string and the document can be substantially identified.

The extraction unit 82 is an example of the extraction means, and extracts words and phrases about a defect of the equipment or parts from the character string created by the preprocessing unit 81. The extraction of words and phrases about the defect in the extraction unit 82 is performed by a match determination with the words and phrases set in the feature amount dictionary 74. That is, the extraction unit 82 extracts words and phrases stored in the feature amount dictionary 74 from the target character string.

AI84 is an abbreviation for Artificial Intelligence, which is a software implementation of the behavior of human intelligence. In the embodiment, AI84 is assumed to be a machine learning model. As a model algorithm, for example, deep learning is used.

AI84 is also an example of selection means, and after machine learning is performed, the words and phrases extracted by the extraction unit 82 are input, and the corresponding characteristic words and phrases are selected. Only one characteristic phrase may be selected, or a plurality of characteristic words may be selected. Further, when a plurality of characteristic words are selected, the sentence may be output as a sentence in which characteristic words are connected. The selection of characteristic words depends on the machine learning described below.

Note that the AI 84 may be constructed so that the extraction unit 82 inputs a phrase or sentence other than the extracted phrase. For example, when the AI 84 has a pre-language processing function, it is conceivable to input the entire character string output by the pre-processing unit 81 to improve the selection accuracy of the characteristic phrase.

Machine learning processing unit 86 performs machine learning of AI84. The machine learning processing unit 86 performs machine learning so that when an appropriate combination of words and phrases included in the feature amount dictionary 74 is given, the corresponding characteristic words and phrases are selected. This allows the AI84 to select a plausible characteristic phrase when given a combination of terms for equipment or component defects, even if the combination is not directly learned. ..

The characteristic phrase acquisition unit 88 acquires the characteristic phrase selected by AI 84 and passes it to the output unit 90.

In the rule-based dictionary setting unit 92, the administrator of the building information processing device 70 inputs and sets the rule-based dictionary 94. The rule-based dictionary 94 is set when the phrase to be acquired is clear. In particular, when acquiring information common to various equipments or parts, it is conceivable to set it in the rule-based dictionary 94 without relying on AI84. Further, the rule-based dictionary 94 is also used when it is desired to directly acquire information from a document input to the document input unit 80 without going through the preprocessing unit 81. A part of the rule-based dictionary 94 can be created by an automatic process, but even in that case, the administrator examines the validity of the rule-based dictionary 94 through the rule-based dictionary setting unit 92. , Set.

Various words and phrases can be set in the rule-based dictionary 94. Examples of settings include those related to building attributes such as building name, total floor area of the building, prefecture where the building is located, and purpose of the building. Further, a phrase indicating the time related to the defect such as the date and time when the defect was inquired and the date and time when the defect was dealt with, or a phrase indicating the details of the defect such as the floor where the defect occurred and the room where the defect occurred may be adopted. As another example, there are words and phrases that represent the person who dealt with the problem, such as the name of the company that dealt with the problem and the name of the person. In general, since there are synonyms or synonyms having similar meanings in words and phrases, synonyms or synonyms are also set in the rule-based dictionary 94.

The rule-based phrase acquisition unit 96 acquires the phrase set in the rule-based dictionary or the phrase corresponding to the phrase from the document acquired by the document input unit 80. That is, the rule-based phrase acquisition unit 96 can acquire a phrase that matches the phrase set in the rule-based dictionary, and the value corresponds to the case where the phrase set in the rule-based dictionary is an item name or the like. You can also get the phrase. The acquired words and phrases are sent to the output unit 90. The output unit 90 is an example of the output means, and outputs the characteristic phrase acquired from the characteristic phrase acquisition unit 88 and the phrase acquired by the rule-based phrase acquisition unit 96 together.

Subsequently, with reference to FIGS. 4 and 5, a document input to the document input unit 80 of the building information processing apparatus 70 will be described. FIG. 4 is an example of a daily business report used in a certain building, and FIG. 5 is an example of a building management correspondence received by telephone.

The daily business report shown in Fig. 4 is a business report prepared daily by workers of a company that performs maintenance management work for a building named "○○ Building". This daily business report is submitted to the XX Building and was prepared on October 15, 2019. The names of day shift workers and night shift workers, the work items performed on that day, and the inspection results are listed. The two thick frames indicated by

reference numerals

110 and 112 in the center of the daily report indicate the details of the defects that occurred on this day and the measures to be taken against them.

In the column of code 110, each item of occurrence date / time, end date / time, responder, content, and action is provided. The date and time of occurrence is 9:52 on October 15, 2019, the date and time of end is 11:13 on the same day, and the number of respondents is 〇〇. The content item is an item corresponding to the "inquiry content", and the requested item that has been contacted is freely described (that is, it is described in a free format) in the corresponding description column. Specifically, there is a request that "There is no water from the back of the women's toilet wash basin on the 14th floor from XX company XX, so there is a response request." In addition, the processing item is an item corresponding to "handling an inquiry", and in the corresponding description column, the items performed by the responder XX are freely described. Specifically, "Status at arrival: Check the situation where water does not come out from the automatic faucet of the corresponding wash basin. Reconnect the power plug of the automatic faucet and perform the reset operation to confirm normal operation. A record of the measures taken with "Confirmation of attendance."

Similarly, in the column of reference numeral 112, matters relating to another defect that 〇〇 corresponded to from 13:35 to 15:21 on the same day are described. Specifically, in the column corresponding to the content item, it is stated that "Mr. △△ company △△ has informed that the doorknob of the doorknob of the entrance door of the exclusive section of the 19th floor △△ ward is rattling." In addition, in the column corresponding to the item of treatment, "Situation at arrival: Confirm that the doorknob of the entrance door of the △△ ward dedicated section on the 19th floor is loose. Tighten the door screw and there is no rattling of the doorknob. It was confirmed that Mr. △△ company △△ witnessed. "

Fig. 5 is a communication document in which the corresponding items are described after being contacted through the call center owned by the maintenance management company. This correspondence only mentions one defect. Specifically, at 13:30, the recipient received a call from the requesting party and was informed about the XX building in Chuo-ku, Tokyo. The received content item corresponds to the "inquiry content", and in the corresponding column, freely enter "◎◎ company ◎◎, 6th floor ◎◎ ward meeting room set temperature until 18:00 25 There is a change request every time. Please respond. "

In addition, the item of cause processing is an item corresponding to "handling of inquiries", and in the corresponding column, "the measured temperature of the corresponding FCU was confirmed to be 26.7 ° C. and changed as follows." The set temperature of the FCU was changed from 27.0 ° C to 25.0 ° C. At 18:00, the set temperature of the corresponding VAV was restored. " This measure was taken by commuter XX on October 18, 2019. The item of the trader name is omitted because it is the same company.

As illustrated in FIGS. 4 and 5, in the maintenance management of a building, a document in a unified format is not always created for contacting or reporting a defect. Different contact or reporting formats may be used for buildings under the same maintenance company, depending on the owner's preference or the historical circumstances of the building. Also, the format of the contact or report may differ depending on the sending tool used by the person making the contact or report. In particular, when cross-cutting processing is performed for a plurality of buildings, it is necessary to obtain a character string from a document having a plurality of formats.

FIG. 6 is a diagram showing the creation of a character string performed by the preprocessing unit 81 of the building information processing apparatus 70. When the documents shown in FIGS. 4 and 5 are input, the preprocessing unit 81 integrates the description of the items corresponding to the "inquiry content" and the "response to the inquiry" to form the "inquiry / response character string". Is being created.

Items

1 and 2 in FIG. 6 are character strings created by extracting from the columns of

reference numerals

110 and 112 in FIG. 4, respectively. Item 3 in FIG. 6 is a character string created by extracting from the document in FIG. It can be said that the character string of each item in FIG. 6 contains information from a request to a countermeasure for one defect.

FIG. 7 is a diagram showing an example of a feature amount dictionary 74 and a characteristic phrase dictionary 78 set in the building information processing apparatus 70. In the example shown in FIG. 7, words and phrases related to defects in building equipment or parts, which are the features set in the feature dictionary 74, are described in the items of the

features

1, 2, and 3. The feature amount 1 is the equipment name, the feature amount 2 is the part name, and the feature amount 3 is set as a descriptive phrase for describing the defect, which indicates a defect identification and a request for dealing with the defect. In addition, characteristic words related to the request set in the characteristic word dictionary 78 are also described.

Items A to F in the left column in FIG. 7 are items set for each defect. For example, the

feature quantities

1, 2, 3 and the characteristic phrase described in item A all describe one defect. Specifically, "air conditioning" and "FCU (fan coil unit)" are set as the equipment name having the feature amount 1, and "VAV (variable air volume method)" and "set temperature" are set as the part names having the feature amount 1. ing. The set temperature is not necessarily a part name, but since it is a main element of air conditioning, it is described in the feature quantity 2 and is treated as a part name. Further, as the description of the defect with the feature amount 3, "operation request", "change request", "hot", and "cold" are described. Then, "temperature change request" is set as a characteristic phrase closely related to these

feature quantities

1, 2, and 3. This characteristic phrase is a summary phrase that simply indicates that a request for change has been made for the temperature regarded as the part name.

In item B in FIG. 7, "air conditioning" and "FCU" are set as the feature amount 1 as in item A. However, there is no defect in item B corresponding to the feature amount 2, and the feature amount 2 is blank. If at least one of the

features

1 and 2 is described, the equipment or part that is the target of the defect can be specified, so that one of the features can be left blank. Further, in the feature amount 3, "abnormal noise", "noise", "offensive odor", "inspection request", and "repair request" are set as defects. The feature amount 3 is always set because it represents a defect. Then, as a characteristic phrase, an "air conditioning inspection request" corresponding to these feature quantities 1 to 3 is set. This characteristic phrase is a summary phrase that briefly indicates that an inspection request has been made to the equipment called air conditioning.

Similarly, in the example shown in FIG. 7, characteristic phrases such as "toilet malfunction", "tube replacement request", "door malfunction", and "other than inquiry (alarm)" are set. The phrase "other than inquiry (alarm)" indicates that an alarm indicating an abnormality has been notified on the central monitoring board. Since there is no person requesting a response in the notification, a description other than an inquiry (alarm) is set. In addition, feature quantities 1 to 3 corresponding to these characteristic terms are also set. In the feature amount 3, for example, words such as "response request" and "repair request" are set in a plurality of items, but the feature amount 3 is special because it represents a defect as a set with the feature amounts 1 and 2. There is no problem.

FIG. 8 is a diagram in which the results of processing each character string shown in FIG. 6 by the extraction unit 82 and AI84 of the building information processing apparatus 70 are added to the table of FIG. The extraction unit 82 extracts all the words and phrases included in the feature amount dictionary 74 from the inquiry / countermeasure character strings indicating each of the defects of items 1 to 3. As a result, in item 1, "automatic faucet" and "wash basin" are extracted as feature amount 1, "power plug" and "water" are extracted as feature amount 2, and "does not come out" and "response request" as feature amount 3. "Is extracted. Similarly, in item 2, "door", "doorknob", "screw", and "rattling" are extracted, and in item 3, "FCU", "VAV", "set temperature", and "change request" are extracted. In the example of FIG. 8, the feature words different from each other are extracted for the items 1 to 3, but the feature words common to a plurality of items may be extracted depending on the contents of the inquiry / coping character string.

In AI84, machine learning is performed using the relationship between the feature amount dictionary 74 and the characteristic phrase dictionary 78 shown in FIG. 7. That is, in AI84, when all the feature amounts 1 to 3 in any of the items A to F of FIG. 7 are extracted and the other feature amounts are not extracted, the corresponding characteristic phrase is selected. Further, in AI84, only a part of the feature amounts 1 to 3 in any of the items A to F in FIG. 7 was extracted, and a part of the feature amounts 1 to 3 included in the other items was also extracted. Even in the situation, it is learned so that appropriate feature words can be selected based on machine learning.

In the example shown in FIG. 8, since the feature quantities 1 to 3 extracted in item 1 are similar to item C shown in FIG. 7, "toilet defect" is selected as the characteristic phrase. Since the feature quantities 1 to 3 extracted in item 2 are similar to item E in FIG. 7, "door defect" is selected as the characteristic phrase. Since the feature quantities 1 to 3 extracted in item 3 are similar to item A in FIG. 7, "temperature change request" is selected as the characteristic phrase. It can be said that the selected characteristic phrase is a classification item for classifying defects.

In the example shown above, the inquiry / response character strings shown in FIG. 6 are classified into the six types of defects of items A to F illustrated in FIG. 7. Further, in the example of FIG. 7, the feature quantities 1 to 3 are only shown as extremely simple examples, and in reality, the number of set words and phrases including synonyms and synonyms also increases. Furthermore, the problems that occur in an actual building are diverse and may be 100 or more or 1000 or more, depending on the degree of subdivision. The inquiry / response character string can easily be 1000 or more, or 10,000 or more, 100,000 or more, depending on the number of buildings and the period for totaling defects. The building information processing apparatus 70 can classify many such character strings at high speed and with high accuracy.

Subsequently, the process based on the rule base will be described with reference to 9 and FIG. FIG. 9 is an example of words and phrases set in the rule-based dictionary 94 of the building information processing apparatus 70. Here, the building name is set as item I, and the prefecture name where the building is located is set as item II.

The rule-based word / phrase acquisition unit 96 of the building information processing device 70 acquires the word / phrase or the corresponding value set in the rule-based dictionary 94 from the document acquired by the document input unit 80. For example, in the daily business report shown in FIG. 4, the phrase "○○ binding" is extracted based on the characters "○○ building middle". However, since there is no description about the prefecture name, the prefecture name is not extracted. Further, in the communication document shown in FIG. 5, since the item of the building name and the item of the address are provided, the words “XX building” and “Tokyo” are acquired from the corresponding description fields. The acquired words and phrases are output by the output unit 90 of the building information processing device 70.

FIG. 10 shows the building name and the prefecture name added to the table of FIG. 8 in addition to the table of FIG. In FIG. 10, in addition to the inquiry / response character string and the feature amounts 1 to 3, characteristic words and phrases output by the output unit 90 and rule-based words and phrases (building name, prefecture) are shown. By using not only characteristic words but also rule-based words, it is possible to extract multifaceted information about building defects.

The extracted information is used in various ways by the PC100. For example, by giving an output for each document as a summary of the document, it can be expected that the document can be organized or the document can be intuitively grasped. In addition, for example, it is useful for summarizing the business of a building in a comprehensive report such as a monthly business report and an annual business report.

Alternatively, for many buildings, statistical examination is possible by aggregating defects for a certain long period (for example, one year). When the statistical values are compared with a specific building, it is possible to analyze the characteristics of defects in the specific building.

In the above explanation, an example of selecting a characteristic phrase for a defect centered on the request shown in FIG. 7 is shown. However, as the characteristic phrase, it is also possible to select a phrase that identifies the defect or a phrase that indicates the content of the countermeasure. In particular, when feature quantities 1 to 3 do not include only requests, and when a characteristic phrase that predicts the content to be dealt with is selected, for example, when instructing a worker to perform work. It will be useful information.

In the above explanation, dealing with defects in building equipment or parts was given as an example. However, this embodiment can be similarly applied to the operation of a building. Here, the operation of a building means the work necessary for using or maintaining the building. The operation of a building includes the operation of equipment or parts of the building (the operation here means the work to continue the use regardless of the occurrence of a defect). In addition, the operation of a building includes the operation of the building itself (business to continue using the building) such as the building or room of the building, or the handling of defects in the building itself.

The above-described embodiment is merely an example of an embodiment, and various embodiments can be taken.

10 Building system, 20, 20a, 20b, 20c Building, 22 Air conditioning equipment, 24 Water supply and drainage equipment, 24a parts, 26 Electrical equipment, 28 elevators, 30 Central monitoring room, 32 Central monitoring panel, 36 Maintenance personnel, 38 Maintenance personnel, 40 Tenant, 42 mobile phone, 44 telephone, 60 communication network, 70 building information processing device, 72 feature quantity dictionary setting unit, 74 feature quantity dictionary, 76 characteristic phrase setting section, 78 characteristic phrase dictionary, 80 document input section, 81 preprocessing Unit, 82 Extraction unit, 84 AI, 86 Machine learning processing unit, 88 Characteristic phrase acquisition unit, 90 Output unit, 92 Rule-based dictionary setting unit, 94 Rule-based dictionary, 96 Rule-based phrase acquisition unit.

Claims

AI that learned to select characteristic terms that represent the characteristics of maintenance management for the building, equipment, or parts by machine learning using words and phrases about the operation of the building or the defects of the equipment or parts of the building.
An extraction means for extracting words and phrases related to the operation of the building or the malfunction of the equipment or parts from the character string in which the contact or report regarding the operation of the building or the malfunction of the equipment or parts of the building is described.
Using the AI, a selection means for selecting the characteristic phrase corresponding to the extracted phrase regarding the operation of the building or the defect of the equipment or part, and
A building information processing device characterized by being equipped with.
In the building information processing apparatus according to claim 1,
Obtain multiple strings from multiple documents in different formats created by users or maintenance managers of the building and describing communications or reports relating to the operation of the building or malfunctions of equipment or parts of the building. Further equipped with pretreatment means to
The extraction means extracts words and phrases about defects of the equipment or parts for each character string.
The building information processing apparatus is characterized in that the selection means uses the AI to select the characteristic phrase for each character string.
In the building information processing apparatus according to claim 1,
Further, the building information processing is provided with a setting means for setting at least a part of words and phrases regarding the operation of the building or the malfunction of the equipment or parts based on the instruction of the manager of the building information processing apparatus. apparatus.
In the building information processing apparatus according to claim 1,
The phrase regarding the operation of the building or the defect of the equipment or part includes the name of the building, equipment or part, and a descriptive phrase describing the operation or defect occurring in the building, equipment or part. A featured building information processing device.
In the building information processing device according to claim 4.
The building information processing apparatus is characterized in that the descriptive phrase is a phrase that describes operation or defect identification, operation or defect response request, or operation or defect response implementation.
In the building information processing apparatus according to claim 1,
The building information processing apparatus is characterized in that the characteristic phrase is a phrase that summarizes the operation or malfunction of the building, equipment, or parts.
In the building information processing apparatus according to claim 1,
The characteristic phrase is a phrase that predicts the operation or defect identification of the building, equipment or part, the request for dealing with the operation or the defect, or the implementation of the countermeasure for the operation or the defect. apparatus.
In the building information processing apparatus according to claim 1,
Further, when the character string or the document from which the character string is acquired contains a predetermined word / phrase, the word / phrase or the word / phrase corresponding to the word / phrase is output together with the characteristic word / phrase for the character string. A building information processing device provided with means.