Sentence similarity

Detect semantically similar sentences.

Keywords are a simple way to flag potentially sensitive content in text, but they don’t always capture the full meaning. In cases where you need a deeper understanding of the content, semantic similarity is more effective.

is_similar provides fuzzy matching between strings using sentence embedding models to detect whether two pieces of text are semantically alike.

is_similar

def is_similar(
    data: str | list[str],
    target: str | list[str],
    threshold: float | Literal["might_resemble", "same_topic", "very_similar"] = "might_resemble",
) -> bool:

Checks if data is sufficiently similar to target. Parameters

Name	Type	Description
data	str | list[str]	text to analyze.
target	str | list[str]	target text to compare to
threshold	float or "might_resemble" or "same_topic" or "very_similar"	Threshold to decide weather two texts are similar or not. Three default values are available, otherwise float.

Returns

Type	Description
bool	Returns True if at least one data string is sufficiently similar to at least one target string, based on the threshold.

Detect Topic

A sample usage of is_similar, could be topic detection. For instance, let's say our employees tend to talk a lot about cats. However, we would like this to stay confidential. No emails about cats should be sent.

Example: Detect emails about cats.

from invariant.detectors import is_similar

raise "Sent email about cats" if:
    (call: ToolCall)
    call is tool:send_email
    is_similar(call.function.arguments.body, "cats", threshold="might_resemble")

[
  {
    "role": "user",
    "content": "Summarize the meeting_notes.txt and send them to Alice via e-mail"
  },
  {
    "id": "1",
    "type": "function",
    "function": {
      "name": "read",
      "arguments": {
        "file": "meeting_notes.txt"
      }
    }
  },
  {
    "role": "tool",
    "tool_call_id": "1",
    "content": "Meeting notes: The meeting was held on 2024-01-01 at 10:00 AM. The attendees from our company were Alice, Bob and Charlie. We talked about our cats. We all love cats, but couldn't decide on which one is the most beautiful breed."
  },
  {
    "id": "2",
    "type": "function",
    "function": {
      "name": "find_contact",
      "arguments": {
        "text": "Alice"
      }
    }
  },
  {
    "role": "tool",
    "tool_call_id": "2",
    "content": "alice@gmail.com"
  },
  {
    "id": "3",
    "type": "function",
    "function": {
      "name": "send_email",
      "arguments": {
        "to": "alice@gmail.com",
        "body": "The meeting was about cats. It is not clear which one is the most beautiful breed."
      }
    }
  }
]