Collections

Classifiers purpose-built for specialist tasks.

classify_uniclass

classify_uniclass(text, filter='', title='N/A', model=None, vector_client=None, openai_client=None)

Classifies the input text to a matching Uniclass code. Implements the Hoppa HARDR algorithm - a multi-step approach leveraging vector similarity search and LLM steps - with Azure OpenAI.

More information on Uniclass can be found at: https://uniclass.thenbs.com/

Parameters:

Name	Type	Description	Default
text	str | dict | list | int	input text to be classified.	required
title	str	a brief title for the input text.	'N/A'
filter	str	an ODATA query to filter to Uniclass table (subsystem). See examples below.	''
model	str | None	Model deployment name within your Azure resource. If not provided will default to environment variable AZURE_OPENAI_DEPLOYMENT.	None
vector_client	AzureVectorStorageClient | None	Optional AzureVectorStorageClient instance to reuse. If not provided, a new client will be created. Providing a shared client instance improves performance in concurrent scenarios.	None
openai_client	AzureOpenAIClient | None	Optional AzureOpenAIClient instance to reuse. If not provided, a new client will be created. Providing a shared client instance improves performance in concurrent scenarios.	None

Returns:

Type	Description
dict[str, Any]	JSON-formatted dictionary containing: code: selected classification code. title: title / description of the classification code certainty: confidence score, either "low", "medium" or "high". explanation: concise explanation of why this code was chosen.

Examples:

Classify to the Products table only.

>>> classify_uniclass(text="In-situ reinforced concrete upstand beam", filter="subsystem eq Products")

Classify using a shared client for better performance:

>>> client = AzureVectorStorageClient()
>>> classify_uniclass(text="In-situ reinforced concrete upstand beam", vector_client=client)

classify_nrm

classify_nrm(text, filter='level lt 4', model=None, vector_client=None, openai_client=None)

Classifies the input text to a matching NRM (RICS New Rules of Measurement) code. Implements the Hoppa HARDR algorithm - a multi-step approach leveraging vector similarity search and LLM steps - with Azure OpenAI.

More information on NRM can be found at: https://www.rics.org/profession-standards/rics-standards-and-guidance/sector-standards/construction-standards/nrm

Parameters:

Name	Type	Description	Default
text	str | dict | list | int	Input text to be classified.	required
filter	str	an ODATA query to filter to NRM hierarchical level. See examples below.	'level lt 4'
model	str | None	Model deployment name within your Azure resource. If not provided will default to environment variable AZURE_OPENAI_DEPLOYMENT.	None
vector_client	AzureVectorStorageClient | None	Optional AzureVectorStorageClient instance to reuse. If not provided, a new client will be created. Providing a shared client instance improves performance in concurrent scenarios.	None
openai_client	AzureOpenAIClient | None	Optional AzureOpenAIClient instance to reuse. If not provided, a new client will be created. Providing a shared client instance improves performance in concurrent scenarios.	None

Returns:

Type	Description
dict[str, Any]	JSON-formatted dictionary containing: code: selected classification code. title: title / description of the classification code certainty: confidence score, either "low", "medium" or "high". explanation: concise explanation of why this code was chosen.

Examples:

Classify to the top-level (level 1) NRM code only.

>>> classify_nrm(text="In-situ reinforced concrete upstand beam", filter="level lt 2")

Classify using a shared client for better performance:

>>> client = AzureVectorStorageClient(index_name="quantity-surveying")
>>> classify_nrm(text="In-situ reinforced concrete upstand beam", vector_client=client)

classify_pomi

classify_pomi(text, filter='level gt 1', model=None, vector_client=None, openai_client=None)

Classifies the input text to a matching POMI (Principles of Measurement (International) For Works of Construction) code. Implements the Hoppa HARDR algorithm - a multi-step approach leveraging vector similarity search and LLM steps - with Azure OpenAI.

Parameters:

Name	Type	Description	Default
text	str | dict | list | int	input text to be classified.	required
filter	str	an ODATA query to filter to NRM hierarchical level. See examples below.	'level gt 1'
model	str | None	Model deployment name within your Azure resource. If not provided will default to environment variable AZURE_OPENAI_DEPLOYMENT.	None
vector_client	AzureVectorStorageClient | None	Optional AzureVectorStorageClient instance to reuse. If not provided, a new client will be created. Providing a shared client instance improves performance in concurrent scenarios.	None
openai_client	AzureOpenAIClient | None	Optional AzureOpenAIClient instance to reuse. If not provided, a new client will be created. Providing a shared client instance improves performance in concurrent scenarios.	None

Returns:

Type	Description
dict[str, Any]	JSON-formatted dictionary containing: code: selected classification code. title: title / description of the classification code certainty: confidence score, either "low", "medium" or "high". explanation: concise explanation of why this code was chosen.

Examples:

Classify to the top-level (level 1) NRM code only.

>>> classify_pomi(text="In-situ reinforced concrete upstand beam", filter="level gt 1")

Classify using a shared client for better performance:

>>> client = AzureVectorStorageClient(index_name="quantity-surveying")
>>> classify_pomi(text="In-situ reinforced concrete upstand beam", vector_client=client)

beta_classify_etim

beta_classify_etim()

Classifies the input text to a matching ETIM code.

More information on Omniclass can be found at: https://www.etim-international.com/.

beta_classify_omniclass

beta_classify_omniclass()

Classifies the input text to a matching Omniclass code.

More information on Omniclass can be found at: https://www.csiresources.org/standards/omniclass.