# US 12,222,689 B2 · Building system with string mapping based on a sequence to sequence neural network

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,222,689 B2
- Document ID: US12222689B2
- Status: Granted
- Kind: B2
- Application number: 18/204,631
- Filed: 2023-06-01
- Published: 2023-12-07
- Granted: 2025-02-11
- Earliest claimed: 2020-05-28
- Assignees: Tyco Fire & Security GmbH
- Applicants: Tyco Fire & Security GmbH
- Inventors: Surajit Borah; Santle Camilus; ZhongYi Jin; Vish Ramamurti; Young M. Lee
- Web page: https://patentgrep.com/patent/US12222689B2

## CPC

- G05B 13/027: {using neural networks only}
- G05B 15/02: electric
- G05B 13/04: involving the use of models or simulators
- G06F 40/295: Named entity recognition
- G06F 40/30: Semantic analysis
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/049: Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G05B 2219/25011: Domotique, I-O bus, home automation, building automation
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A building system including one or more memory devices configured to store instructions that, when executed by one or more processors, cause the one or more processors to receive training data including acronym strings and tag strings, train a sequence to sequence neural network based on the training data, receive an acronym string for labeling, the acronym string comprising a particular plurality of acronyms, and generate a tag string for the acronym string with the sequence to sequence neural network, wherein the sequence to sequence neural network outputs a tag of the tag string for one acronym of the particular plurality of acronyms based on the one acronym and contextual information of the acronym string, wherein the contextual information includes other acronyms of the particular plurality of acronyms.

## Claims

### Claim 1

A building system comprising one or more memory devices configured to store instructions thereon that, when executed by one or more processors, cause the one or more processors to:
receive training data comprising a plurality of tag strings, wherein a tag string of the plurality of tag strings is a translation of an acronym string of a plurality of acronym strings, and wherein the plurality of acronym strings include a plurality of acronyms that represent a plurality of entities of a building;
train a sequence to sequence neural network based on the training data;
receive an acronym string comprising a particular plurality of acronyms; and
output, using the sequence to sequence neural network, a tag of a tag string for one acronym of the particular plurality of acronyms based on a placement of the one acronym within the acronym string.

### Claim 2 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to:
receive, from a user via a user device, a selection of one or more first acronym strings of the plurality of acronym strings;
train the sequence to sequence neural network based on the one or more first acronym strings of the plurality of acronym strings; and
determine, responsive to training the sequence to sequence neural network, labels for one or more second acronym strings of the plurality of acronym strings.

### Claim 3 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to train the sequence to sequence neural network by:
identifying one or more first acronym strings of the plurality of acronym strings that are similar to one or more second acronym strings of the plurality of acronym strings;
selecting, from at least one of the one or more first acronym strings or the one or more second acronym strings, a given acronym string to represent the one or more first acronym strings and the one or more second acronym strings; and
training the sequence to sequence neural network based on the given acronym string.

### Claim 4 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to:
identify, based on characteristics of the plurality of acronym strings, a given acronym string to represent one or more acronym strings of the plurality of acronym strings;
display, via a display of a user device, the given acronym string; and
receive, from a user via the user device, a translation of the given acronym string.

### Claim 5 (depends on 4)

The building system of claim 4, wherein the sequence to sequence neural network is trained based on the translation of the given acronym string.

### Claim 6 (depends on 1)

The building system of claim 1, wherein the sequence to sequence neural network is a long-short term memory (LSTM) sequence to sequence neural network.

### Claim 7 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to:
determine a number of strings of the particular plurality of acronym strings;
compare the number of strings to a threshold level; and
select the sequence to sequence neural network to translate the acronym string from a group of available models in response to a determination that the number of strings is greater than the threshold level.

### Claim 8 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to receive the training data from a database, and wherein the training data is based on data of one or more buildings and the acronym string is associated with the building.

### Claim 9 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to:
remove at least one of spaces or special characters from the plurality of acronym strings;
apply a segmentation model to the plurality of acronym strings to identify the plurality of acronyms;
generate an acronym vocabulary by removing redundant acronyms from the plurality of acronyms; and
train the sequence to sequence neural network based on the training data and the acronym vocabulary.

### Claim 10 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to:
receive a set of acronym strings for the building for translation;
select the plurality of acronym strings from the set of acronym strings;
receive the plurality of tag strings from a user device, each tag string of the plurality of tag strings being a translation of respective acronym strings of the plurality of acronym strings;
train the sequence to sequence neural network based on the training data; and
translate remaining acronym strings of the set of acronym strings with the sequence to sequence neural network.

### Claim 11 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to receive a manual selection of the plurality of acronym strings from a user device.

### Claim 12 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to receive the plurality of tag strings from a user device via user input provided by a user via the user device, the user input indicating tag translations of particular acronyms of the plurality of acronym strings.

### Claim 13 (depends on 1)

The building system of claim 1, wherein the instructions cause the one or more processors to cause a user device to display the plurality of acronym strings to a user for manual translation.

### Claim 14

A method comprising:
receiving, by one or more processing circuits, training data comprising a plurality of tag strings, wherein a tag string of the plurality of tag strings is a translation of an acronym string of a plurality of acronym strings, and wherein the plurality of acronym strings include a plurality of acronyms that represent a plurality of entities of a building;
training, by the one or more processing circuits, a sequence to sequence neural network based on the training data;
receiving, by the one or more processing circuits, an acronym string comprising a particular plurality of acronyms; and
outputting, by the one or more processing circuits using the sequence to sequence neural network, a tag of a tag string for one acronym of the particular plurality of acronyms based on a placement of the one acronym within the acronym string.

### Claim 15 (depends on 14)

The method of claim 14, further comprising:
receiving, by the one or more processing circuits, from a user via a user device, a selection of one or more first acronym strings of the plurality of acronym strings;
training, by the one or more processing circuits, the sequence to sequence neural network based on the one or more first acronym strings of the plurality of acronym strings; and
determining, by the one or more processing circuits, responsive to training the sequence to sequence neural network, labels for one or more second acronym strings of the plurality of acronym strings.

### Claim 16 (depends on 14)

The method of claim 14, wherein training the sequence to sequence neural network includes:
identifying, by the one or more processing circuits, one or more first acronym strings of the plurality of acronym strings that are similar to one or more second acronym strings of the plurality of acronym strings;
selecting, by the one or more processing circuits, from at least one of the one or more first acronym strings or the one or more second acronym strings, a given acronym string to represent the one or more first acronym strings and the one or more second acronym strings; and
training, by the one or more processing circuits, the sequence to sequence neural network based on the given acronym string.

### Claim 17 (depends on 14)

The method of claim 14, further comprising:
identifying, by the one or more processing circuits, based on characteristics of the plurality of acronym strings, a given acronym string to represent one or more acronym strings of the plurality of acronym strings;
displaying, by the one or more processing circuits via a display of a user device, the given acronym string; and
receiving, by the one or more processing circuits, from a user via the user device, a translation of the given acronym string.

### Claim 18

One or more non-transitory storage medium configured to store instructions thereon that, when executed by one or more processors, cause the one or more processors to:
receive training data comprising a plurality of tag strings, wherein a tag string of the plurality of tag strings is a translation of an acronym string of a plurality of acronym strings, and wherein the plurality of acronym strings include a plurality of acronyms that represent a plurality of entities of a building;
train a sequence to sequence neural network based on the training data;
receive an acronym string comprising a particular plurality of acronyms; and
output, using the sequence to sequence neural network, a tag of a tag string for one acronym of the particular plurality of acronyms based on a placement of the one acronym within the acronym string.

### Claim 19 (depends on 18)

The one or more non-transitory storage medium of claim 18, wherein the instructions cause the one or more processors to:
receive, from a user via a user device, a selection of one or more first acronym strings of the plurality of acronym strings;
train the sequence to sequence neural network based on the one or more first acronym strings of the plurality of acronym strings; and
determine, responsive to training the sequence to sequence neural network, labels for one or more second acronym strings of the plurality of acronym strings.

### Claim 20 (depends on 18)

The one or more non-transitory storage medium of claim 18, wherein the instructions cause the one or more processors to train the sequence to sequence neural network by:
identifying one or more first acronym strings of the plurality of acronym strings that are similar to one or more second acronym strings of the plurality of acronym strings;
selecting, from at least one of the one or more first acronym strings or the one or more second acronym strings, a given acronym string to represent the one or more first acronym strings and the one or more second acronym strings; and
training the sequence to sequence neural network based on the given acronym string.