Meng Lin

Xing Wu, Guangyuan Ma, Meng Lin et al.

Dense passage retrieval aims to retrieve the relevant passages of a query from a large corpus based on dense representations (i.e., vectors) of the query and the passages. Recent studies have explored improving pre-trained language models to boost dense retrieval performance. This paper proposes CoT-MAE (ConTextual Masked Auto-Encoder), a simple yet effective generative pre-training method for dense passage retrieval. CoT-MAE employs an asymmetric encoder-decoder architecture that learns to compress the sentence semantics into a dense vector through self-supervised and context-supervised masked auto-encoding. Precisely, self-supervised masked auto-encoding learns to model the semantics of the tokens inside a text span, and context-supervised masked auto-encoding learns to model the semantical correlation between the text spans. We conduct experiments on large-scale passage retrieval benchmarks and show considerable improvements over strong baselines, demonstrating the high efficiency of CoT-MAE. Our code is available at https://github.com/caskcsg/ir/tree/main/cotmae.

Huipeng Ma, Luan Zhang, Dandan Song et al.

In multi-hop reasoning, multi-round retrieval-augmented generation (RAG) methods typically rely on LLM-generated content as the retrieval query. However, these approaches are inherently vulnerable to knowledge overshadowing - a phenomenon where critical information is overshadowed during generation. As a result, the LLM-generated content may be incomplete or inaccurate, leading to irrelevant retrieval and causing error accumulation during the iteration process. To address this challenge, we propose ActiShade, which detects and activates overshadowed knowledge to guide large language models (LLMs) in multi-hop reasoning. Specifically, ActiShade iteratively detects the overshadowed keyphrase in the given query, retrieves documents relevant to both the query and the overshadowed keyphrase, and generates a new query based on the retrieved documents to guide the next-round iteration. By supplementing the overshadowed knowledge during the formulation of next-round queries while minimizing the introduction of irrelevant noise, ActiShade reduces the error accumulation caused by knowledge overshadowing. Extensive experiments show that ActiShade outperforms existing methods across multiple datasets and LLMs.

Xing Wu, Chaochen Gao, Meng Lin et al.

Before entering the neural network, a token is generally converted to the corresponding one-hot representation, which is a discrete distribution of the vocabulary. Smoothed representation is the probability of candidate tokens obtained from a pre-trained masked language model, which can be seen as a more informative substitution to the one-hot representation. We propose an efficient data augmentation method, termed text smoothing, by converting a sentence from its one-hot representation to a controllable smoothed representation. We evaluate text smoothing on different benchmarks in a low-resource regime. Experimental results show that text smoothing outperforms various mainstream data augmentation methods by a substantial margin. Moreover, text smoothing can be combined with those data augmentation methods to achieve better performance.

Yoshiyuki Fujikawa, Hengyi Li, Xuebin Yue et al.

Oracle bone inscriptions (OBIs) contain some of the oldest characters in the world and were used in China about 3000 years ago. As an ancient form of literature, OBIs store a lot of information that can help us understand the world history, character evaluations, and more. However, as OBIs were found only discovered about 120 years ago, few studies have described them, and the aging process has made the inscriptions less legible. Hence, automatic character detection and recognition has become an important issue. This paper aims to design a online OBI recognition system for helping preservation and organization the cultural heritage. We evaluated two deep learning models for OBI recognition, and have designed an API that can be accessed online for OBI recognition. In the first stage, you only look once (YOLO) is applied for detecting and recognizing OBIs. However, not all of the OBIs can be detected correctly by YOLO, so we next utilize MobileNet to recognize the undetected OBIs by manually cropping the undetected OBI in the image. MobileNet is used for this second stage of recognition as our evaluation of ten state-of-the-art models showed that it is the best network for OBI recognition due to its superior performance in terms of accuracy, loss and time consumption. We installed our system on an application programming interface (API) and opened it for OBI detection and recognition.