LT6273B - System and method of ofaviation parts structure recognition in structure free text - Google Patents

Abstract

The invention relates to system and method of a text processing and more particularly to electronic text data processing system and a method with integrated artificial neuron networks. The system comprises at least one computer with digital data storage, where stored commands of computer program to recognize text of aviation parts at least the number, quantity, condition in the structure free text.Mentioned computer program includes a word processing commands and network commands, and processor means for process the computer program commands stored in the digital storage. The invention allows automatically get the result - the structured data, such as aeronautical engineering parts list to a specific part number, quantity and condition from any type of text input such as the e-mail, or other digital format documents. It can be used to make process automatic - scan e-mails or other text format document and integrate this data into the aviation companies for search of used parts inventory, ERP and other systems.

Description

1

FIELD OF THE INVENTION The present invention relates to a text processing system and method, and more specifically to an electronic text data processing system and a method for an integrated trained artificial neural network.

Technical level

In the aerospace market, every detail has its own unique number that identifies them. This allows you to simplify communication between manufacturers, buyers and sellers, avoiding errors due to complex details. All current aviation parts search and ERP systems (business management systems) use structured detail details (part number, quantity, status). Most buyers and sellers interact by e-mail, sending data, such as text in e-mail content or attached media, usually without a structure or structure specific to one letter or attached document. In order to find the required information about the required aeronautical details in a particular e-mail or other document, relevant data, such as the name of the aeronautical component, should be entered into the aeronautical component search or ERP system. This is done by transcribing or copying and pasting the textual parts of the e-mail content into a specific aeronautical component search or ERP system. Relevant manual operations, especially when dealing with large amounts of data, require significant human resources. U.S. Patent No. 8392409 B1 discloses a system and method for analyzing and reviewing e-mails, wherein displaying information related to e-mail includes receiving information relating to multiple e-mails as a response to a search query provided. A computer program product stored on a computer-readable medium includes a code for retrieving search information and a code for the categorization of e-mails according to the intended categories, a user code for the tag containing the tag, and a tag linking with at least one of a plurality of e-mail linking codes. The drawback of the present invention is that the system performs a search in the text of an e-mail only according to the words t.y. Search engines require specific search terms that are searched for by many processed emails. 2

U.S. Pat. No. 6,273,848 B1 discloses a method for evaluating the content of an electronic message text. Semantic search operations are performed using message content, where the identified message is converted to a tagged stream, which is recognized by each of the identified tag related commands that are subsequently executed. Message content analysis and evaluation is carried out using keywords where the system and method are limited to the search of the targeted keywords in the messages. U.S. Pat. US 11 / 397,593 discloses a system and method for checking non-systemic messages and correspondence that may be received in the form of an e-mail or telephone call statement. The extracted text is scanned to provide useful and relevant information using a list of words and phrases where these words and phrases can be described as industry-recognized words and phrases. The present invention is essentially limited to the search for exact terms in the text of an electronic message.

The closest to the invention is the "Amber Artificial Intelligence" and "Amber A.I." Plugin " products for search in Locatory.com, for aerospace identification using database matching. These products are limited to work only with Locatory.com search engine. The invention allows automated retrieval of the result from any type of text input with data on searched aeronautical details, such as electronic mail or other format electronic documents, such as a list of aeronautical parts with a particular part number, number and condition. This can be used to automate the scanning of e-mails or other forms of text documents and to integrate these data into search engine, ERP and other systems used by aviation companies. In this way, the productivity of the employees is significantly increased and at the same time the need for employees is reduced, since the invention automatically reads and lists the searched aircraft parts in an automated form from an e-mail or other document, i. the user no longer needs to search for a specific detail name and manually create a list of them.

BRIEF DESCRIPTION OF THE INVENTION The invention relates to a system for identifying aeronautical details in text-free text and searching in a structure. The system includes at least 3

EN 6273 B one computer with electronic data storage medium, stored in computer program commands for recognizing at least numbers, quantities, states of at least parts of the structure in a text without structure, where said computer program includes text processing commands and trained artificial neural network commands, and processor tool for executing computer program commands stored on an electronic data carrier. The system also includes at least one computer device associated with said at least one computer with data carriers containing information about the aviation part numbers, their quantities and their states. The invention also encompasses a method for recognizing aviation parts in a non-structured text using a system comprising at least a computerized electronic data medium, storing computer program commands for recognizing at least numbers, quantities, states of at least parts of the structure in a text without structure, wherein said computer program includes said computer program text processing commands and artificial neural network commands, and a processor tool for executing computer program commands stored on an electronic data carrier. The method includes training the artificial neural network in the electronic text to find at least the parts of the aeronautical parts, the amounts thereof, the information corresponding to the state, the use of the trained artificial neural network in order to obtain the text of the non-structural electronic format using the program in which said artificial neural network is integrated in a processor, text processing, and retrieving, retrieving, and storing the necessary information, such as number, quantity, or condition of aviation in the medium. Also, using a computer program is a comparison of aeronautical part numbers, quantities or their state information with the details of aircraft parts, quantities or their state on the media of other computer devices.

A brief description of the drawings

FIG. 1 Principal scheme of system identification of parts, numbers and states of aeronautical parts in non-systematic text;

FIG. 2 Schematic of a system for searching for details of number, quantity and condition information.

FIG. 3 Neural Network Training Scheme 4

LT 6273 B

FIG. 4 Text processing scheme;

DETAILED DESCRIPTION OF THE INVENTION The invention relates to a method and a search engine for identifying information, numbers and states of aviation details in a non-systematic text.

The system comprises at least one first computer (PR) connected to at least a computerized electronic data carrier (1), which contains a stored computer program (2) whose commands are designed to identify at least numbers, quantities, and statuses of the aviation parts in a text without structure (3) , wherein said computer program (2) comprises text processing commands and commands for a trained artificial neural network, and comprising a processor (6) for executing computer program commands stored in an electronic data medium (1). The system also includes an electronic text information retrieval tool (7) connected to said first computer (PR) via a Computer Communication Networks (CCP), such as email servers associated with computer programs that can be associated with a computer program (2). , at least other computer tools (CCPs) connected to data storage media (DSL) containing information about the aviation components, their quantities and their states. The at least one first computer (PR) is connected to said at least other computer means (CCPs) via means of computer communication networks (CCPs) so that the program (2) executed by the processor (6) can obtain information from said at least other computer media (DSL) and can compare it with the stored, received, input, textual information, number and condition information of at least one first computer (PR) (1). The method of recognizing information in an unstructured text relating to the numbers, quantities, and states of aeronautical parts using said system includes selecting an artificial neural network (21) and placing it on a computer medium (1); an artificial neural network training (22) comprising processing of information related to the details in a computerized medium (1) in the details of the details, which may be at least a part number, a number of parts and a condition, comprising translating each word into a set of numeric parameters; receiving an e-mail (23); executing a text information processing program (2) (24); electronic message content recognition (25); performing a component search (26) and retrieving search results 5

EN 6273 B (27).

By training (22) an artificial neural network during an ongoing translation (28), the number of characters comprising a group of letters, consecutive numbers, characters from an ASCII table, characters outside the ASCII table, @, dashes is checked in the given word and what is the length of the word. The training also uses a similar amount of words that are not aeronautical parts numbers and all words. About 70% of the data in the detail database can be used for the training of the artificial neural network, while the rest of the data from the database is used to determine the accuracy of the trained artificial neural network separating the unknown data, i. data not used during training; the simultaneous execution of an artificial neural network and a word processing program (2) in a computer medium (1) in a processor device (6), so that the said program (2), the text information input to the database stored in the computer medium (7), is in the text information Identify (30) The parts, numbers and states of the aeronautical parts, where the details of the aforementioned part numbers, their quantities and status are stored on a computer medium (1).

The above word processing program is executed (24) in the processor device (6) so that the entered (23) textual information is processed according to the following steps: from the input text, any set of characters, is ejected (31) into the part number structure-like characters compounds such as dates, aircraft names, phone numbers, addresses; tabs, such as various non-standard spaces, punctuation marks, are replaced (32) by the word space symbol; the input text, any set of characters, is broken (33, 34) into rows, such as the new line symbol, the rows of the HTML table, etc., and these are words by space character; iterations are performed on each row and through the words of the line where the words are converted (35) to a set of parameters, in the same way as for the training of the artificial neural network; this set is transmitted to a program using a trained artificial neural network, which returns (36) the numerical probability of -1 to 1, so that the word is a part of the aviation part. If the probability is close enough to 100%, for example about 90%, the word is considered as a detail and is investigated by the line containing that word.

After finding (37) the aviation part numbers, the part quantity (6) 6 is further searched

LT 6273 B t.y. in the part number environment, such as the line where the part number was found, the number is searched for if several numbers are found - the last one is taken. If this information is found, it is checked that it is not a numbering of lines, where the quantity is compared to the previous line and if it increases by one, it is considered to be numbering, and whether the quantity was not supported by a separate part number, for example 100 may be and quantity and part number, so part numbers are discarded from the line before searching for quantity.

After finding (38) the quantities of aviation components, the information (39) and the state of the part are further searched (39). Initially, the status keywords are translated into their abbreviations, and then the line contains the following shortcuts that can be selected from the group consisting of, but not limited to: FN - directly from the factory, NO - new equipment, NS - new surplus, OH - after capital repairs, RP - corrected, SV - used, AR - stage as dismantled, NA - not set.

In this way, the entire input text is processed, a list (40) is created with the data found and stored on a computer medium. This list has an exact structure that includes at least the part numbers, their states, and the quantities that can be presented in any format, such as JSON, which makes it possible to integrate with any existing or emerging search engine, ERP or other work with aviation components system for hinged interfaces.

The exemplary embodiment of the present invention should not be construed as the sole example of implementation and it should be understood that various modifications and adaptations of the system and method are readily apparent to those skilled in the art upon the basis of the teachings provided.

Claims (1)

7 EN 6273 B Definition 1. A system comprising a computer program product stored on a computer-readable medium for the processing of e-mails, the selection of content information and the conduct of searches in computer-accessible databases, is characterized by the fact that a computer program is stored on a system's electronic data medium (1) (2) includes word processing commands and trained artificial neural network commands for recognizing aerospace part numbers, quantities and states in a text without structure (3). 2. A method for processing computer e-mails, selecting content information, and searching for computer-accessible databases in a computer-readable computer program comprises the steps of: a) selecting an artificial neural network (21) and placing it on a computer medium (1) ); b) training an artificial neural network (22) comprising processing of information related to the details in a real-time data base in a computer medium (1), which may be at least a part number, a number of parts and a condition, comprising translating each word into a set of numeric parameters; c) receiving (23) an electronic message on a computer medium (1); d) executing (24) a text processing program (2) for processing textual information of the e-mail message by means of a processor; (e) the identification (25) of the content of the electronic mail relating to the numbering, quantity and condition of the aeronautical parts in an unstructured text; f) performing a component search (26) according to the e-mail content recognized in step e); g) storing the results of step f) in a computer medium (1) and output (27) as an information display device. δ LT 6273 B 3. A method according to claim 2, characterized in that the training of the neural network comprises the steps of: a) a number of details of the details of the details in the computer medium (1), which may be at least a part number and status, processing comprising the translation of each word (28) into a set of numeric parameters and the number of characters to be checked, comprising a group of letters, a number of consecutive digits, characters from an ASCII table, characters outside the ASCII table, @, dashes is in a given word and the length of the word; b) repeating step (a) (29) with all words in the text; (c) simultaneous processing of the artificial word neural network (1) and the word processing program (2) in the computer medium (2) in the processor (6) for said program (2); The text information entered into the text information input to the database stored in the computer medium (7) would recognize (30) the numbers of the aeronautical parts, their present quantities and the states in which said details of numbers, their quantities and status are stored on a computer medium (1). d) repeating step c) with all part numbers found. 4. Method according to claim 2 or 3, characterized in that said word processing program (4) is executed in a processor (6) to process the entered textual information according to the following steps: a) from the input text, any character set , (31) character combinations similar to the part number structure are discarded, but not part numbers; b) tabs, such as various non-standard spaces, punctuation marks, are replaced (32) by the word space symbol; c) the input text, any set of characters, is broken (33) into rows, and the words (34) are separated by a space character; d) iterating (35) each line and the words of the line where the words are converted into a set of parameters; e) the set obtained in step d) is transmitted to a program using a trained artificial neural network, which returns (36) a numeric probability of -1 to 1, so that the word g LT 6273 B is the aviation part number; f) further detailing (37) the amount of detail; g) the amount of detail found in step f) is checked (38) whether it is a numbering of rows where the amount is compared with the amount of the previous row; h) searching for information (39) detailing the state of the part, translating the state keywords into their abbreviations, and searching in the line for these abbreviations that can be selected from the group consisting of, but not limited to: FN - directly from the factory, NO - new equipment, NS - new surplus, OH - after major repairs, RP-corrected, SV - used, AR - stage as dismantled, NA - not established. i) generating (40) a list of accurate structures comprising at least part numbers, their states and quantities, and stored on a computer medium.