Contradiction analysis

class expert.core.contradiction.contradiction_analysis.ContradictionDetector(transcription_path: str | PathLike, video_path: str | PathLike, lang: str = 'en', device: torch.device | None = None, output_dir: str | PathLike = None, chosen_intervals: List[int] = None, interval_duration: int = 60)[source]

Bases: object

Class for working with detector of contradiction.

Parameters

  • transcription_path (str | PathLike) – Path to JSON file with words from transcribation.

  • video_path (str | PathLike) – Path to local video file.

  • lang (str, optional) – Speech language for text processing [‘ru’, ‘en’]. Defaults to ‘en’.

  • device (str, optional) – Device type on local machine (GPU recommended). Defaults to None.

  • output_dir (str | PathLike) – Path to save JSON with information about contradictions.

  • chosen_intervals (list[int], optional) – Flags for intervals for analyzis. Defaults to []. For default settings will be analyzed full text.

  • interval_duration (int, optional) – The length of intervals (value in seconds). Defaults to 60.

Returns

Path to the contradiction report.

Return type

str

Raises

NotImplementedError – If ‘lang’ is not equal to ‘en’ or ‘ru’.

get_utterances(all_words: List[dict]) → List[str][source]

Function for getting utterances for comparison. Split text with speech pauses :param all_words: Fragment’s words to create utterances :type all_words: List[dict]

Returns

Utterances for comparison

Return type

List[str]

property device: torch.device

Check the device type.

Returns

Device type on local machine.

Return type

torch.device

get_contradiction(entered_text: str) → str[source]

Function for text analyzing. Creates json file with predictions.

Parameters

entered_text (str) – Utterance for analysis.