Better machine translation: having a computer able to translate languages with the quality of human experts has always been a challenge.With deep learning, it is possible to train advanced relevance ranking models from past interactions/judgments to rank documents for a given query (both represented as numerical vectors). Providing the most useful results first in the ranked list is fundamental. Learning to rank: currently, the vast majority of search engines identify a set of candidate documents from the corpus of information (matching) and order them by relevance to satisfy the user information need (ranking).Using large pre-trained models which are finely tuned for your use case (potentially using transfer learning techniques), helps to build the foundation in advanced multimedia retrieval, reducing the effort of continuously supervised metadata tagging.
Better image/video representations: extracting semantic features from images and videos (such as the objects and entities involved rather than just pixels and colour-related information).Improvements in this field could lead to completely new types of Information Retrieval systems that behave like human experts: the systems won’t just return a list of documents to satisfy your information need but also synthesize a comprehensive natural language response backed by supporting evidence (documents). Generating text can be useful in many Information Retrieval areas, such as query auto-completion, query spellchecking, document summarization, search results explainability (summarizing the information that the document contributes to the user information need), and more. Text generation: language modeling techniques flourished and reached mainstream news thanks to outstanding results in generating text that is almost indistinguishable from human-made.Better text representations: moving away from the bag-of-words model (where terms are sequences of characters) to a multi-dimensional numerical approach (vectorized), which is able to model terms as semantic units of information linked to each other with meaning.Too many to discuss them all here, for sure! But here are some examples for you to understand: There is, of course, a subfield of ML that helps search engineers tackle these problems as well as bring new and interesting capabilities to search engines: deep learning.Īpplying deep learning techniques to solve search problems is often called Neural Search.ĭeep learning helps us solve the problems we explained before, but what are some other things that deep learning can contribute to search? We learned before about Machine Learning, which is a subfield of Artificial Intelligence. Keep reading to find out a couple of resources to further enhance your knowledge at the end! In learning, they are used to learn relevance ranking functions, classify query intent and documents, and offer personalized results.vector-based), to identify meaning, synonyms, relations between terms and concepts, and for spellchecking. In knowledge representation, they are used to build better data structures and search algorithms (e.g.In image/video recognition, they are used to extract features and search a multimedia of corpus information.In natural language processing (NLP), they are used to better understand and model the user information need and corpus of information, and for text segmentation to target specific passages of information.So what problems do both AI and ML help solve in Information Retrieval? Here are some examples:
0 kommentar(er)
