Google's open-source ARTEMIS framework enables natural-language-driven Android automation with dual Flash/Pro execution modes, MCP integration for AI IDEs, and a three-layer fallback location strategy, achieving 99% success on AndroidWorld benchmarks while supporting CI/CD and real-device farms.
Stanford's Christopher Manning argues that while LLMs excel at linguistic form, they still falter at pragmatics, discourse, and dialogue — revealing linguistics' irreplaceable role in probing whether models truly understand meaning beyond pattern abstraction.
The article explains that NL2SQL translates natural language into SQL queries, offering a narrow but reliable form of data‑centric programming, while highlighting its boundaries, recent multi‑step extensions, and its role within broader natural‑language‑to‑code ecosystems.
NL2SQL transforms natural language queries into executable SQL, enabling non‑technical users to retrieve data by simply speaking, and the article explains its workflow, evolution from rule‑based to large‑model approaches, current performance on the Spider benchmark, remaining challenges, and real‑world use cases.
An AI‑driven review of 168 ICML 2026 oral papers reveals that only 105 could be fully reproduced, with a median replication cost of $8,900, many hidden flaws, and 903 blind‑spot issues that human reviewers missed, questioning the trustworthiness of top‑conference publications.
The MLNLP 2026 online symposium on July 26 will feature leading AI researchers from Shanghai Jiao Tong University, CUHK (Shenzhen) and Tencent Hunyuan presenting talks on lifelong learning, generative model fine‑tuning, and unified multimodal reinforcement learning, with registration now open.
This article reviews the evolution of deep learning, its key concepts, model architectures, training strategies, and applications—especially in speech, vision, and natural language processing—highlighting seminal research, comparative analyses, and current challenges for future AI development.
The article introduces Natural Language Processing as an AI subfield, outlines its four main task categories—classification to sequence, sequence to classification, synchronous and asynchronous seq‑to‑seq—and details the typical preprocessing pipeline including corpus collection, cleaning, tokenization, stemming, lemmatization, POS tagging, NER, and chunking.
This article provides concise answers to frequent Natural Language Processing questions, covering the distinction between NLP and NLG, popular pretrained models, deep‑learning architectures, word‑vector techniques, named‑entity recognition, sentiment, semantic and syntax analysis, part‑of‑speech tagging, language models, core tasks, real‑world applications, challenges, future trends, interpretability, and essential tools and libraries.
In RAG applications, ambiguous user queries often hinder retrieval effectiveness, so rewriting queries before search—through normalization, synonym expansion, linguistic rules, LLM‑based generation, query decomposition, and multi‑view strategies—can improve relevance, but must avoid over‑expansion, semantic drift, and added latency.
This article introduces the basics of Natural Language Processing, covering core tasks such as language understanding and generation, common applications, essential linguistic analyses, environment setup with Python libraries, hands‑on code examples for preprocessing, POS tagging, NER, sentiment analysis using both classical and transformer models, text generation with GPT‑2, and discusses challenges and Rust‑centric integration strategies.
A 30B‑A3B model, trained with reverse‑perplexity supervised fine‑tuning, two‑stage reinforcement learning, and a multi‑round generate‑verify‑revise inference loop, achieves gold‑medal performance on IMO, USAMO and IPhO contests using over 100 k token natural‑language reasoning without external tools.
Tokens are the smallest language building blocks that AI models process, representing characters, words, subwords, punctuation or emojis; they determine context window size and generation speed, so tokenization directly impacts model understanding accuracy and efficiency, as explained in the 2026 Token Report.
Analyzing five years of ICLR submission metadata reveals that titles containing a question mark boost acceptance to 45.5% in 2022, while emerging keywords such as diffusion, sparse, and planning dominate high‑acceptance lists, and older topics like federated learning, adversarial attacks, and security suffer low acceptance and high withdrawal rates.
The article explains why conventional Retrieval‑Augmented Generation often produces incorrect answers despite retrieving relevant documents, introduces the Disco‑RAG framework that adds a structured reading step using argument trees and relation graphs, and shows how this three‑step approach dramatically improves performance on long‑document and ambiguous‑question benchmarks without any model training.
The talk examines why large language models, despite impressive performance on knowledge‑based tasks, struggle dramatically when required to learn new information from the immediate input context, analyzes systematic biases behind this limitation, and explores rubric‑based synthesis as a potential remedy.
This article explains the self‑attention mechanism behind transformers, detailing its core components, mathematical formulation, step‑by‑step example, multi‑head extension, industry use cases, and a thorough comparison with RNN and CNN approaches, all supported by concrete numbers and citations.
Text2DSL lets users describe search requirements in plain language, uses DeepSeek to generate Elasticsearch DSL, validates the DSL locally with Elasticsearch/Easysearch, iteratively refines it up to five times, and achieves over 95% first‑try accuracy while cutting query‑building time by at least threefold.
Leveraging AI and natural language processing, Tmall’s quality assurance team transformed traditional manual testing into a semi‑automated and fully automated pipeline—covering requirement analysis, test case generation, data construction, execution, and validation—resulting in significant efficiency gains, traceability, and continuous improvement across multiple business lines.
This article explains how Retrieval‑Augmented Generation (RAG) combines vector databases with large language models to let non‑technical users ask natural‑language questions and receive precise SQL statements, detailing the workflow, architecture, chunking methods, performance gains, and remaining challenges.
This article surveys the most influential large language models released since 2017, detailing the core innovations of Transformer, BERT, GPT series, T5, Retrieval‑Augmented Generation, and the latest LLaMA and Meta models, while highlighting their architectures, training paradigms, and impact on NLP research.
This article explains how recurrent neural networks handle variable-length sequences, why LSTM's gated memory cells solve long-term dependency issues, and how attention mechanisms improve translation by weighting encoder hidden states, laying the groundwork for Transformer-based LLMs.
This article explains how Baidu Map’s AI navigation system converts spoken queries into accurate map instructions by combining speech recognition, intent parsing, large‑language‑model reasoning, tool calling, and memory‑reflection techniques, showcasing the underlying technologies that enable instant, context‑aware responses.
This article explains how Baidu Map’s AI voice assistant converts spoken commands into precise navigation actions by detailing the speech‑to‑text pipeline, intent parsing, template and generative approaches, tool‑calling mechanisms, memory and reflection capabilities, and future directions for intelligent agents.
Tracing the evolution of large language models—from early bag‑of‑words techniques, through word embeddings, RNNs, attention mechanisms, Transformers, BERT, and GPT—this article explains each breakthrough, its limitations, and how they culminated in ChatGPT’s conversational AI.
This comprehensive guide walks you through the Transformer architecture—from its encoder‑decoder structure and self‑attention mechanism to multi‑head attention, positional embeddings, and practical PyTorch implementations—providing clear visualizations and code examples for deep learning practitioners.
This comprehensive article examines Retrieval‑Augmented Generation (RAG), covering its historical evolution, core theory, implementation variants, practical code examples, diverse applications, current controversies, and future research directions within the AI and NLP landscape.
Pre‑training enables AI models to first acquire a universal knowledge map from massive unlabelled text, then quickly adapt to specific tasks with minimal labelled data, offering superior generalization, reduced annotation costs, and versatile applications across chatbots, content creation, retrieval, coding assistance, and more.
The paper investigates how a Quantum Convolutional Neural Network (QCNN) can classify complex bank policy documents, compares it with a classical TextCNN on 50‑ and 100‑sample scenarios, and shows QCNN achieves up to 7‑12% higher accuracy and F1 scores, reducing compliance risk.
The article presents an in‑depth overview of DeepInsight Copilot, an AI‑driven business intelligence product that streamlines data, information, insight, and decision‑recommendation stages, detailing its functional modules, intelligent agents, multi‑generation technical evolution, architecture, model fine‑tuning, and future challenges and solutions in data analysis.
This article explores the application of Large Language Models (LLM) in text information detection and extraction, focusing on blue-collar recruitment data processing. It details the implementation of LLM through prompt engineering, RAG enhancement, and model fine-tuning to improve data cleaning efficiency and accuracy.
Prompt engineering uses in‑context learning to turn large language models into precise, task‑aware assistants by providing well‑crafted prompts that guide the model’s probability distribution, reduce hallucinations, and unlock hidden knowledge without any parameter tuning.
This article explains how AI-powered digital humans can modernize banking by offering modular, multi‑modal interaction, personalized multilingual service, 24‑hour availability, and risk‑aware automation, while detailing the underlying AI foundation, decision engine, visual rendering, and deployment strategies.
This article explains how language models evolved from character‑level embeddings to word‑level and finally to sub‑word tokenizers, highlighting the efficiency, vocabulary coverage, and practical engineering challenges of sub‑word segmentation in modern AI systems.
The authors present a semantic‑graph‑enhanced uncertainty modeling framework that captures token, sentence, and paragraph dependencies, propagates uncertainty through entity relations and contradiction probabilities, and achieves roughly a 20 % gain in paragraph‑level hallucination detection on WikiBio and NoteSum compared with existing uncertainty‑based baselines.
This tutorial explains how to integrate Spring AI's Model Context Protocol (MCP) with Chat2DB to enable secure, natural‑language queries against a PostgreSQL database, covering configuration, core Java code, execution flow, and sample SQL queries.
This article provides a comprehensive overview of LLM-based Text-to-SQL technology, covering its background, evolution, challenges, various LLM-driven methods, benchmark datasets, evaluation metrics, and future research directions to guide researchers and practitioners in advancing natural language interfaces for databases.
This article examines how an AI‑powered essay‑grading platform combines PaddleOCR and Baidu's Wenxin large model to automate scoring, generate personalized feedback, and reduce teachers' grading workload by over six times, while improving student learning outcomes across hundreds of Chinese schools.
This article explains what AI Transformers are, why they matter, their key components and mechanisms, various applications ranging from language processing to bioinformatics, and how they differ from traditional neural networks, providing a comprehensive overview of Transformer architecture and its impact on modern AI research.
The article examines Hugging Face's ERobot, an AI model that leverages Transformer-based pre‑trained models to execute a wide range of automation tasks through natural‑language commands, discusses its technical foundations, real‑world applications, future prospects, and the challenges it must overcome.
Effective prompt design for large language models requires clear goals, relevant context, explicit input/output formats, evaluation criteria, and illustrative examples, combined with specific language, step‑by‑step instructions, edge‑case handling, ethical considerations, and proper tokenization, encoding, decoding, and post‑processing to produce accurate, concise, low‑hallucination responses.
This article examines the challenges of low parcel matching rates in JD Express logistics and proposes a large‑model‑based intelligent identification system, detailing its architecture, accuracy validation, cost‑saving cache strategy, and future prospects for improved efficiency and personalized services.
This article details the end‑to‑end development of an in‑vehicle voice assistant, covering motivation, functional design, technology stack selection, dialogue flow, privacy, third‑party integration, wake‑word detection, on‑device speech recognition, noise filtering, NLP processing, and deployment considerations.
The paper describes Baidu’s AI‑native advertising platform that employs a multi‑agent architecture built on large‑language models—combining large‑small model collaboration, domain SOP‑driven coordination, and long‑term memory—to enable natural‑language understanding, proactive planning, execution and human‑like responses, illustrated by GBI analytics and JarvisBot operations, delivering higher consumption, accuracy, speed and efficiency.
This article examines the adoption of large language models for intelligent advertising copy creation, detailing business challenges, model selection criteria, training data preparation, fine‑tuning methods, performance evaluation, deployment results, while highlighting the trade‑offs between model size, cost, and output quality.
This article explains how Transformer models work by detailing the encoder‑decoder structure, self‑attention, multi‑head attention, positional encoding, and feed‑forward networks, and shows their applications in machine translation, recommendation systems, and large language models.
This article uses a playful dialogue to demystify how large language models like GPT‑4 work, covering data collection, vectorization, the transformer’s attention mechanism, position encoding, training stages, multilingual translation, reasoning puzzles, and alignment, all illustrated through the tale of a curious learner named Wuming.
This article explains how large language models work, why well‑crafted prompts are essential, and presents practical strategies—such as clarity, conciseness, focus, role‑setting, delimiters, few‑shot examples, and step‑by‑step instructions—to help users obtain accurate and relevant responses from AI systems.
Postdoctoral researcher Fan Yuankai from Fudan University will present his work on enabling natural-language queries for database systems, covering NL2SQL approaches, reliable ranking mechanisms, and guiding large models to generate accurate SQL, aiming to improve usability for users unfamiliar with query languages.
Bilibili combats personal‑attack and trolling comments by combining sector‑specific keyword databases, user‑group analysis, advanced word‑matching (including pinyin and homophone detection) and multiple NLP/graph models, which has cut personal‑attack reports in entertainment, film and gaming by about 32 % and trolling reports by roughly 25 % between June and December 2023.
This article shares the practical application of large language models within Zhihu’s internal Jianqiao analytics platform, covering business background, knowledge taxonomy organization, natural‑language‑to‑filter conversion, natural‑language data analysis, and summarizing challenges, solutions, and future outlooks.
This article reviews the historical development of artificial intelligence, explains how AI technologies such as large language models are reshaping data processing and analysis, and discusses the practical challenges, trust issues, and governance requirements when applying AI to the data industry.
The X-STA algorithm, introduced by Alibaba Cloud’s PAI and researchers from South China University of Technology, leverages gradient‑decomposed knowledge sharing, teacher‑guided attention, and multi‑level alignment to enhance cross‑lingual machine reading comprehension, achieving state‑of‑the‑art results on three multilingual MRC benchmarks.
This article explores Chinese word segmentation, illustrating its linguistic nuances with a humorous example, explains key methods—including dictionary‑based, statistical, and deep‑learning approaches—and provides Python code using a simple dictionary algorithm and the popular jieba library to demonstrate practical implementation.
This article introduces finite‑state transducers, explains their mathematical definition, illustrates state‑transition examples such as binary counters, word detection, and parentheses matching, explores key applications in speech synthesis, spell‑checking, lemmatization, transliteration, and lexical analysis, and provides a concise Python implementation.
This article details how Zhihu's internal Bridge platform integrates large language models for business analysis, knowledge taxonomy, natural‑language‑to‑filter conversion, and ad‑hoc data queries, describing the workflow, technical hurdles, iterative improvements, and future directions.
This article explains the Word2Vec algorithm, its CBOW and Skip‑Gram architectures, cosine similarity mathematics, training process with negative sampling, and provides a concise Python example using the gensim library.
A joint Stanford and UC Berkeley study reveals that GPT‑4’s performance on mathematics, code generation, and visual‑reasoning tasks sharply declined between March and June 2023, with accuracy dropping from 97.6% to 2.4% on a prime‑checking benchmark and executable code rates falling from 52% to 10%.
This article documents a step‑by‑step case study of applying GPT‑4 to automatically review and filter game dialogue, detailing the initial challenges, iterative prompt refinements, added safety rules, and the resulting improvements compared with earlier models.
This article introduces Chat2DB, an open‑source AI‑enhanced multi‑database client, walks through its download, installation, configuration, demonstrates natural‑language‑to‑SQL, SQL explanation and optimization features using sample MySQL tables, and outlines future roadmap for developers.
This article explains prompt engineering for large language models like ChatGPT, covering its definition, essential techniques such as diverse prompting strategies, problem restatement, background provision, gradient prompting, example inclusion, role‑playing, and the importance of systematic experimentation and quantitative evaluation to achieve high‑quality, task‑specific AI outputs.
This article explains why crafting good prompts matters more than memorizing facts, introduces prompt engineering for large language models, and provides practical techniques—such as diverse prompting strategies, problem restatement, background provision, gradient prompting, role‑playing, and systematic evaluation—to help users get the most out of ChatGPT and similar AI systems.
This guide thoroughly explains the major decoding strategies for neural text generation in HuggingFace Transformers—including greedy, beam, diverse beam, sampling, top‑k, top‑p, sample‑and‑rank, beam sampling, and group beam search—detailing their principles, Python implementations with LogitsProcessor components, workflow diagrams, comparative analysis, and references to original research.
This article provides a comprehensive analysis of the Transformer model behind ChatGPT, covering its origin, core mechanisms such as embedding, positional encoding, self‑attention, multi‑head attention, a step‑by‑step translation example, and the broader implications for AI research and industry.
This article translates Ibrahim John’s book on prompt engineering for ChatGPT, explaining a wide range of prompting techniques—including instruction, role, seed‑word, zero‑shot, few‑shot, and reinforcement‑learning prompts—through clear English descriptions, formulas, and illustrative examples to help readers obtain high‑quality model outputs.
In a live talk, AI veteran Wu Jun explains why the hype around ChatGPT is overblown, traces the history of language models from the 1970s, details the massive compute and data requirements, and discusses the real impact of large‑scale AI on society and work.
This comprehensive tutorial demonstrates how to leverage the official ChatGPT API and advanced prompt engineering techniques to develop a command-line tool that translates natural language instructions into precise shell commands, covering iterative prompt design, constraint implementation, and Node.js integration.
The article examines how ChatGPT automates many manual tasks, reducing demand for translators, editors, customer support reps, and data analysts, while boosting opportunities for chatbot developers, NLP engineers, data scientists, content creators, and software developers, and outlines broader industry advancements driven by AI.
The article reviews GPT‑4’s expanded visual and coding abilities, modest arithmetic gains, and capacity to use external tools, while highlighting persistent shortcomings in planning, long‑range context, complex calculations, and societal risks such as misinformation, concluding that despite impressive advances it remains far from true artificial general intelligence.
Fudan University's Natural Language Processing Lab introduced MOSS, the country's first conversational large language model capable of dialogue generation, programming, factual QA and ethical reasoning, with plans for open‑source release despite current limitations in Chinese language proficiency.
MOSS, a Chinese large‑language model released by Fudan University, offers ChatGPT‑like functions such as text generation, summarization, translation, and code writing, while being open‑source and free during preview, yet it still lags behind due to limited data, compute, and model size.
This article presents an in‑depth overview of Tencent's TexSmart natural‑language understanding system, highlighting its fine‑grained NER capabilities, knowledge‑base combination methods, remote‑supervision via similar entities, multi‑source zero‑shot fusion, experimental results, and practical insights from a recent NLP summit.
This article provides a comprehensive, plain‑language walkthrough of the 2017 “Attention Is All You Need” paper, explaining the Transformer’s architecture, core mechanisms such as embedding, positional encoding and self‑attention, and discussing its broader impact on AI research and applications.
The article presents a detailed interview between Time magazine and ChatGPT, exploring how large language models work, their ethical risks, limitations, and potential societal impact while offering guidance on responsible use and future expectations.
This article presents a comprehensive overview of multimodal pre‑training, describing its motivation, architecture choices, large‑scale Chinese image‑text dataset construction, training optimizations, performance benchmarks, downstream applications, and a Q&A session that highlights practical deployment considerations.
At EMNLP 2022, Alibaba Cloud’s PAI platform showcased three pioneering few-shot NLP papers—KECP, SpanProto, and UPT—demonstrating advanced prompt-tuning techniques, knowledge-enhanced models, and a unified learning paradigm that push the boundaries of low-resource language understanding.
This article traces the historical convergence of mathematics and linguistics, from 19th‑century pioneers to post‑war computer‑driven research, highlighting how statistical, probabilistic, and formal methods laid the foundation for machine translation, morphological analysis, and contemporary natural language processing.
This article presents a comprehensive retrieval‑based dialogue system designed to assist customer‑service agents by recommending candidate replies, detailing its five‑layer architecture, metric suite, text and vector retrieval modules, ranking strategies, and real‑world deployment results across multiple business scenarios.
This paper presents an end‑to‑end approach for extracting relational triples directly from speech signals, bypassing intermediate transcription, and demonstrates its effectiveness on synthesized speech versions of the CoNLL04 and TACRED datasets, highlighting challenges such as length constraints and cross‑modal alignment.
This article explains the end‑to‑end pipeline of query understanding for e‑commerce search, covering preprocessing, segmentation, spell correction, normalization, and expansion, and discusses both academic research and industry implementations with examples and references.
This article reviews the integration of contrastive learning into text generation, explains why it helps mitigate exposure bias, introduces the CoNT framework with three key improvements, presents extensive experiments on translation, summarization, code comment and data‑to‑text tasks, and discusses practical deployment considerations.
This article introduces paper-digest, an AI‑driven assistant that parses full research papers to generate concise, clear summaries, helping students read more efficiently, researchers share findings faster, and science communicators produce effective overviews.
The article explains Augmented Analysis—using AI‑driven natural‑language queries, intelligent assistants, and automated business insights—to enable non‑technical users to explore data, gain actionable recommendations, and boost business value, illustrated with real‑world use cases and practical guidance for embedding the technology into everyday workflows.
This article explains how the Fine‑ and Coarse‑Granularity Hybrid Self‑Attention (FCA) mechanism reduces BERT’s computational cost by over 50% while keeping accuracy loss under 1%, detailing the method, experimental results, and its significance for efficient large‑scale language models.
The NL4Opt competition at NeurIPS 2022, co‑hosted by Huawei Cloud and two Canadian universities, introduces the first natural‑language‑based mathematical‑modeling dataset, offering named‑entity‑recognition and model‑generation tracks, a detailed schedule, and prize incentives to advance AI‑powered optimization research.
The article highlights ten representative Meituan technical papers accepted at ACM SIGIR 2022, spanning personalized opinion tagging, cross‑domain sentiment classification, dialogue summarization transfer, universal retrieval, CTR prediction, image behavior modeling, and topic segmentation, each summarized with abstracts and download links for researchers.
This article presents the JiuGe system developed by THUNLP for automatically generating Chinese classical poetry, detailing its research motivations, model architecture—including salient‑clue, working‑memory, topic‑memory, style‑transfer and reinforcement‑learning components—implementation, applications, and future directions.
This article reviews contemporary text summarization methods, covering extractive approaches such as TextRank and clustering, abstractive models like Seq2Seq with attention, pointer‑generator networks, and recent pre‑trained transformers including BART, CPT and PEGASUS, highlighting their strengths, limitations, and combined strategies.
The paper reviews classic summarization models, then proposes DSMRC‑S, a span-level extractive dialogue summarization method using distant supervision and a machine‑reading‑comprehension framework, with token‑level labeling and density‑based span selection, achieving state‑of‑the‑art BLEU and ROUGE improvements on a large Meituan dialogue dataset.
WantWords, an open‑source reverse dictionary from Tsinghua University, lets users input a description and receive matching words across Chinese and English, leveraging a multi‑channel model from a AAAI‑20 paper and offering customizable part‑of‑speech and rhyme options.
This article provides a comprehensive overview of entity linking, detailing its definition, the two-stage pipeline of entity recognition and disambiguation, common methodologies such as candidate generation and ranking, advanced approaches, challenges like unlinkable mentions, and various applications in knowledge graphs, text mining, and question answering.
This article explains how treating negative response candidates with varying difficulty through a hierarchical curriculum learning framework—combining corpus‑level and instance‑level curricula—enhances dialogue response selection models, backed by experiments on Douban, Ubuntu, and E‑Commerce datasets.
Seq2Path treats each sentiment tuple as an independent tree path, training with average path loss and decoding via constrained beam search with a discriminative token, achieving state‑of‑the‑art results on five aspect‑based sentiment analysis datasets and deployment in Alibaba Entertainment AI Brain.
This article explores how AI and natural language processing technologies are applied to music lyric creation, covering background challenges, rhyme retrieval methods, advanced language models such as SongNet, decoding strategies, style transfer, and a multi‑level generation platform that aims to streamline professional songwriting.
This presentation reviews the concepts, tasks, and challenges of information extraction from unstructured text, covering closed and open settings, relation extraction, joint extraction, and open extraction methods, and discusses recent advances such as segment‑attention, global‑rationale models, ETL, TPLinker, and maximal‑clique based approaches with experimental results.
This talk presents a series of graph‑based techniques for natural language processing, including the Story Forest system for hot event discovery, the GIANT framework for ontology creation and user interest modeling, and a divide‑and‑conquer approach to long‑text matching that leverages graph neural networks and community detection.
Lightning Flash 0.3 expands the PyTorch Lightning ecosystem with eight new computer‑vision and NLP tasks, modular API design, integrated model hubs, visualisation callbacks, customizable data‑source hooks, and a central registry for model backbones, all illustrated with concrete code examples.
The paper introduces a hybrid curriculum learning framework that tackles emotion shift and confusing labels in emotion recognition in conversation by applying nested curriculum stages at both conversation and utterance levels, enabling a progressive easy‑to‑hard training that markedly boosts classic ERC model performance across four public datasets and is already deployed in Alibaba’s entertainment AI brain script health‑check service.
This article reviews the fundamentals of knowledge graphs, their practical uses in question answering, search and recommendation, and surveys recent research on entity linking—including dual‑encoder retrieval, BERT‑based models, multilingual approaches, and zero‑shot methods—while also outlining modern knowledge‑graph construction pipelines and open challenges.
The IOSPE system, which uses BERT‑based scoring, transformer‑pointer generation, and personalized distribution to automatically extract and generate selling points for millions of e‑commerce products, earned the AAAI 2022 Artificial Intelligence Innovation Application Award and has boosted click‑through rates and user dwell time across JD.com platforms.
