This article surveys and organizes research works in a new paradigm in natural language processing, which we dub “prompt-based learning.” Unlike traditional supervised learning, which trains a model to take in an input x and predict an output y as P ( y|x ), prompt-based learning is based on language models that model the probability of text directly. To use these models to perform prediction tasks, the original input x is modified using a template into a textual string prompt x′ that has some unfilled slots, and then the language model is used to probabilistically fill the unfilled information to obtain a final string x̂ , from which the final output y can be derived. This framework is powerful and attractive for a number of reasons: It allows the language model to be pre-trained on massive amounts of raw text, and by defining a new prompting function the model is able to perform few-shot or even zero-shot learning, adapting to new scenarios with few or no labeled data. In this article, we introduce the basics of this promising paradigm, describe a unified set of mathematical notations that can cover a wide variety of existing work, and organize existing work along several dimensions, e.g., the choice of pre-trained language models, prompts, and tuning strategies. To make the field more accessible to interested beginners, we not only make a systematic review of existing works and a highly structured typology of prompt-based concepts but also release other resources, e.g., a website NLPedia–Pretrain including constantly updated survey and paperlist.

Conversational robots have been widely used in the medical field to monitor, diagnose, and manage patient diseases. The construction of question and answer systems relies on domain knowledge, and according to research, there are currently no conversational bots available to answer postoperative thyroid eye disease questions. To alleviate healthcare resources, we have developed a multi-agent dialogue system for answering post-operative thyroid eye disease questions. The system consists of three agents: an intent understanding agent, a knowledge processing agent, and an intelligent interaction support agent, each responsible for intent recognition, knowledge base building, and conversation generation. The system is scalable and reusable with a division of labour between the agents working together to complete the dialogue tasks. In the experimental part, we used the dialogue system to simulate a post-operative thyroid eye surgery dialogue scenario, and the system was able to accurately recognise intent and answer the user’s questions well.

Knowledge Graphs (KGs) store information in the form of (head, predicate, tail)-triples. To augment KGs with new knowledge, researchers proposed models for KG Completion (KGC) tasks such as link prediction; i.e., answering (h; p; ?) or (?; p; t) queries. Such models are usually evaluated with averaged metrics on a held-out test set. While useful for tracking progress, averaged single-score metrics cannot reveal what exactly a model has learned -- or failed to learn. To address this issue, we propose KGxBoard: an interactive framework for performing fine-grained evaluation on meaningful subsets of the data, each of which tests individual and interpretable capabilities of a KGC model. In our experiments, we highlight the findings that we discovered with the use of KGxBoard, which would have been impossible to detect with standard averaged single-score metrics.

As AI systems engage in extended thinking processes for research tasks, interaction becomes increasingly important for effective long-horizon research. In these tasks, many current deep research systems still follow an “input–wait–output” paradigm: users submit a query and receive results only after a non-interruptible process with limited visibility into intermediate decisions. In multi-agent systems, delayed feedback can amplify early errors, limit query refinement during investigation and reduce opportunities for expertise integration. To address these limitations, we introduce Deep Cognition, a multi-agent deep research system that enables users to monitor and steer an ongoing research process through targeted interventions during execution. Deep Cognition provides two key capabilities: (1) Transparent, controllable, and interruptible interaction that reveals AI reasoning, exposing intermediate plans and evidence and allowing users to intervene at intermediate checkpoints; (2) Fine-grained bidirectional dialogue that enables per-subagent control during execution.