AI_FIX_GRAMMAR

Overview

AI_FIX_GRAMMAR is an AI grammar correction function provided by Singdata Lakehouse. It automatically fixes grammar, spelling, and punctuation errors in input text. Supports Chinese, English, Japanese, French, and other languages, and can intelligently unify mixed-language text. If the input text has no grammar errors, the function returns the original text unchanged. One line of SQL handles text correction.

Singdata pushes AI computation down to the storage and execution engine layer. Data is processed intelligently within the platform without leaving the system, ensuring data security while significantly reducing task latency.

Syntax

AI_FIX_GRAMMAR( <model>, <content> )

Parameters

Required Parameters

model

Specifies the model for grammar correction. Supports two sources:

Source 1: API Gateway Endpoint (Recommended)

A platform administrator pre-configures model services in the API Gateway. Regular users reference them with the endpoint: prefix, without needing to know the underlying connection details.

'endpoint:<endpoint_name>' -- Examples 'endpoint:qwen3-max-preview' 'endpoint:qwen3.5-plus'

Source 2: API Connection Object

Users create their own connection objects via CREATE API CONNECTION, suitable for custom service addresses, authentication keys, or private deployment models.

-- Create a connection object CREATE API CONNECTION conn_bailian TYPE ai_function PROVIDER = 'bailian' BASE_URL = 'https://dashscope.aliyuncs.com/api/v1' API_KEY = 'sk-xxxxxxxxxxxxxxxxxxxxxxxx'; -- Reference using <connection_name>:<model_name> format SELECT AI_FIX_GRAMMAR('conn_bailian:qwen3.5-plus', 'He dont know what to did.');

CREATE API CONNECTION field descriptions:

Field	Description
TYPE	Fixed as ai_function
PROVIDER	Model provider identifier, e.g. 'bailian', 'openai', 'anthropic'
BASE_URL	Base API URL of the model service
API_KEY	Authentication key for calling the service

content

The input text to correct, type STRING (supports CHAR/VARCHAR/STRING). Supports Chinese, English, Japanese, French, and other languages — no need to specify the language manually; the model detects it automatically.

Return Value

STRING type — the corrected text. If the input text has no grammar errors, the original text is returned unchanged.

Error Behavior

By default, if the function cannot process the input, it returns NULL without raising an error. In multi-row queries, rows that error return NULL without affecting other rows.

Usage Notes

• NULL returns NULL, empty string returns empty string: When content is NULL, returns NULL; when content is '', returns '' — no error in either case. Whitespace-only strings (e.g. ' ') are sent to the model and typically return an empty string.
• Error-free text is not modified: If the input text is grammatically correct, the function returns the original text without unnecessary changes.
• Filter before correcting: For large tables, use WHERE content IS NOT NULL AND LENGTH(content) > 0 to filter empty values first, reducing unnecessary model calls.
• Be aware of semantic changes: AI_FIX_GRAMMAR targets grammar correction, but in rare cases (e.g. when the original text has logical contradictions) the model may alter meaning. For semantically sensitive scenarios, spot-check the corrected results.
• Mixed-language text unification: For user-generated content mixing languages, AI_FIX_GRAMMAR intelligently unifies to the dominant language — useful for content standardization.
• Combine with other AI functions: Use AI_FIX_GRAMMAR to clean text before AI_SENTIMENT or AI_SUMMARIZE to improve downstream analysis quality.
• Results are non-deterministic: LLM-based corrections may vary slightly across model versions or call times; the same input may produce different results across executions.

Examples

Basic Usage

-- English grammar correction SELECT AI_FIX_GRAMMAR( 'endpoint:qwen3-max-preview', 'He dont know what to did.' ) AS fixed; -- Returns: He doesn't know what to do. -- Chinese grammar correction SELECT AI_FIX_GRAMMAR( 'endpoint:qwen3-max-preview', '我昨天去了北京，那里的天气很好的很。' ) AS fixed; -- Returns: 我昨天去了北京，那里的天气很好。 -- No errors — text returned unchanged SELECT AI_FIX_GRAMMAR( 'endpoint:qwen3-max-preview', 'The quick brown fox jumps over the lazy dog.' ) AS fixed; -- Returns: The quick brown fox jumps over the lazy dog.

Multilingual

-- Japanese (remove duplicate past tense) SELECT AI_FIX_GRAMMAR( 'endpoint:qwen3-max-preview', '私は昨日学校に行きましたでした。' ) AS fixed; -- Returns: 私は昨日学校に行きました。 -- French (fix tense + contraction) SELECT AI_FIX_GRAMMAR( 'endpoint:qwen3-max-preview', 'Je suis allé au magasin et je achète du pain.' ) AS fixed; -- Returns: Je suis allé au magasin et j''ai acheté du pain. -- Mixed Chinese-English text unification SELECT AI_FIX_GRAMMAR( 'endpoint:qwen3-max-preview', '这个product的quality真的很good，我very喜欢。' ) AS fixed; -- Returns: 这个产品的质量真的很好，我非常喜欢。

Semantic Understanding Capabilities

-- Subject-verb agreement SELECT AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', 'She go to school every days.') AS fixed; -- Returns: She goes to school every day. -- Tense error SELECT AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', 'Yesterday I go to the store and buy some food.') AS fixed; -- Returns: Yesterday I went to the store and bought some food. -- Chinese redundant wording SELECT AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', '他大约大概有三十岁左右。') AS fixed; -- Returns: 他大约三十岁。 -- Chinese missing sentence component SELECT AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', '通过这次活动，使我受到了教育。') AS fixed; -- Returns: 这次活动使我受到了教育。

Batch Processing Table Data

SELECT id, original_review, AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', original_review) AS cleaned_review FROM customer_reviews WHERE original_review IS NOT NULL AND LENGTH(original_review) > 0;

Combining with Other AI Functions

-- Fix grammar first, then analyze sentiment SELECT id, AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', content) AS fixed_content, AI_SENTIMENT('endpoint:qwen3-max-preview', AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', content) ) AS sentiment FROM feedback WHERE content IS NOT NULL; -- Fix grammar first, then summarize SELECT id, AI_SUMMARIZE('endpoint:qwen3-max-preview', AI_FIX_GRAMMAR('endpoint:qwen3-max-preview', content), 30 ) AS summary FROM articles WHERE content IS NOT NULL;

Limitations

• model parameter is required: Omitting it causes the error AI function must have at least two arguments.
• Invalid model format causes an error: model must use 'endpoint:<name>' or '<connection_name>:<model_name>' format; incorrect format causes Invalid model coordinates.
• Correction scope: Focuses on grammar, spelling, and punctuation; does not guarantee correction of all collocation issues.
• Semantic preservation: In rare cases (when the original text has logical contradictions), meaning may be altered; spot-check semantically sensitive scenarios.
• Input length is model-limited: Input text length is limited by the underlying model's context window; overly long text may be truncated.
• Model dependency: Requires a configured Endpoint in the AI Gateway. For available Endpoints, contact your platform administrator or check Lakehouse Studio → AI → Model Management.