Abstract: Implementations of the present disclosure provide a method, a device, and an apparatus for verifying a veracity of a statement, and a medium. In One method, includes: training data including a statement, an evidence set, and a label are acquired. The statement is divided into a plurality of phrases based on a grammatical analysis of the statement. A phrase verification model is trained based on the training data and the plurality of phrases, so that the phrase verification model determines a plurality of phrase veracities of the plurality of phrases respectively based on the evidence set. A statement verification model is trained based on the training data and the plurality of phrases, so that the statement verification model determines a statement veracity of the statement based on the evidence set, where the plurality of phrase veracities provide an interpretation for the statement veracity.

Abstract: Embodiments of the present disclosure relate to a method, an apparatus, an electronic device and a medium for event extraction. The method comprises: extracting a plurality of named entities from a document as a plurality of event arguments. The method further includes determining an event type and a template corresponding to the event type in the document. The method also includes filling the plurality of event arguments in respective locations in the template to generate a plurality of candidate event records, and filtering the plurality of candidate event records to obtain one or more target event record. In this way, respective candidate event records are generated through iteration while event extraction is performed at a document level, which can avoid performance fluctuations caused by manual selection of an event role generation sequence, and can also avoid under-fitting brought about by parallel generation to thus improve the event extraction accuracy.

Abstract: A method of consistency detection of a document and an abstract is proposed. The method includes determining a first sample and first annotation information, the first annotation information indicating that a first abstract and a first document included in the first sample are inconsistent, at least one of a plurality of text elements in the first abstract being labeled as inconsistent with the first document; generating a first adversarial sample by applying interference information to the first sample, the interference information being applied to the first sample and other text elements in the first abstract except for the at least one text element; and training, at least based on the first sample, the first adversarial sample, and the first annotation information, a consistency detection model according to a training objective. The obtained trained model can better detect and track a part of the abstract that are inconsistent with the document.

Abstract: A text chain generation method includes selecting a to-be-matched phrase chain from a phrase chain set to match the initial phrase chain and determining the largest common subsequence between the to-be-matched phrase chain and the initial phrase chain; updating the initial phrase chain by adding a word from the to-be-matched phrase chain and other than the largest common subsequence into the initial phrase chain; using the updated initial phrase chain as a new initial phrase chain and repeating the previous steps until traversing all phrase chains in the phrase chain set to obtain an updated phrase chain; and connecting a left node located in each branch of the updated phrase chain and not connected to any node to a preset common start node and connecting a right node located in each branch of the updated phrase chain and not connected to any node to a preset common end node.

Abstract: Provided are an advertisement title rewriting method, apparatus and device and a storage medium. The method includes steps described below. Preset information in an advertisement title is replaced with a corresponding placeholder so that the advertisement title is converted into a title template; a rewritten title template is generated through a preset neural network model according to the title template, an advertisement attribute key-value table and a query cluster; and the placeholder in the rewritten title template is replaced with the corresponding preset information so that the rewritten title template is converted into a rewritten title.