+41 544 36 40
  • en
  • de
  • annotation

    An annotation is a short note or comment added to a text, image or another document. In linguistics, this is mostly manual extraction of certain features of natural language in texts. Example: determining the gender of proper names by marking individual names in texts with labels such as female, diverse, neutral, etc.

     

    Annotation and linguistics

    Annotation is a technique used in linguistics to mark various features within texts. A classic annotation is morphological annotation, in which word types (e.g. nouns, verbs, adjectives) are marked. Syntactic annotation is also widely used. This marks the syntactic roles of words.

     

    Automated annotation using machine learning models

    In the past, annotation had to be done manually by humans. Nowadays, there are several techniques that allow annotation to be performed by machines. In recent years, the use of trained ML models has increased, either to support annotators or to replace them completely. Both methods have their advantages and disadvantages. Manual annotation is considered to be more accurate, but is also time consuming and expensive. Automatic annotation can be faster, but is less accurate and may require additional review by a human annotator.

    Annotated corpora (collections of text) can also be used to train your own language models. This is achieved by training the model with many exemplary texts and then being able to predict the part of speech of a word based on its context and structure.

     

    Where are annotations used?

    Annotations play an important role in corpus linguistics. In corpus linguistics, large collections of texts (corpora) are used for research. Annotated corpora are a valuable resource for linguists because they allow the study of various linguistic phenomena, e.g. word frequencies, statistically frequent occurrences of word combinations (collocations) or syntactic structures. The annotation of these corpora makes it possible to identify certain linguistic patterns. These can be used to inform linguistic theories and improve our understanding of how language is used in real contexts.

    Another important application of annotation is computational linguistics, which is concerned with modelling language. In computational linguistics, annotations are used to generate training data for machine learning. These models can then be used for tasks such as language translation, text summarisation and sentiment analysis.

     

    To summarise:

    Annotation is a powerful tool in linguistics that makes it easier to identify and understand linguistic features and patterns. It is used in various fields such as corpus linguistics, computational linguistics and natural language processing, and plays a crucial role in the development and evaluation of language models. The availability of high quality data is essential for the further development of language models.

     

    Sources:

    https://www.klippa.com/de/blog/informativ/was-ist-datenannotation/ (German)

    https://www.clickworker.de/ki-glossar/datenannotation/ (German)

    https://medium.com/@orendaminds/what-is-annotation-in-ai-ml-df63af8599e8