No Arabic abstract
Pre-trained model such as BERT has been proved to be an effective tool for dealing with Information Retrieval (IR) problems. Due to its inspiring performance, it has been widely used to tackle with real-world IR problems such as document ranking. Recently, researchers have found that selecting hard rather than random negative samples would be beneficial for fine-tuning pre-trained models on ranking tasks. However, it remains elusive how to leverage hard negative samples in a principled way. To address the aforementioned issues, we propose a fine-tuning strategy for document ranking, namely Self-Involvement Ranker (SIR), to dynamically select hard negative samples to construct high-quality semantic space for training a high-quality ranking model. Specifically, SIR consists of sequential compressors implemented with pre-trained models. Front compressor selects hard negative samples for rear compressor. Moreover, SIR leverages supervisory signal to adaptively adjust semantic space of negative samples. Finally, supervisory signal in rear compressor is computed based on condition probability and thus can control sample dynamic and further enhance the model performance. SIR is a lightweight and general framework for pre-trained models, which simplifies the ranking process in industry practice. We test our proposed solution on MS MARCO with document ranking setting, and the results show that SIR can significantly improve the ranking performance of various pre-trained models. Moreover, our method became the new SOTA model anonymously on MS MARCO Document ranking leaderboard in May 2021.
Similar question retrieval is a core task in community-based question answering (CQA) services. To balance the effectiveness and efficiency, the question retrieval system is typically implemented as multi-stage rankers: The first-stage ranker aims to recall potentially relevant questions from a large repository, and the latter stages attempt to re-rank the retrieved results. Most existing works on question retrieval mainly focused on the re-ranking stages, leaving the first-stage ranker to some traditional term-based methods. However, term-based methods often suffer from the vocabulary mismatch problem, especially on short texts, which may block the re-rankers from relevant questions at the very beginning. An alternative is to employ embedding-based methods for the first-stage ranker, which compress texts into dense vectors to enhance the semantic matching. However, these methods often lose the discriminative power as term-based methods, thus introduce noise during retrieval and hurt the recall performance. In this work, we aim to tackle the dilemma of the first-stage ranker, and propose a discriminative semantic ranker, namely DenseTrans, for high-recall retrieval. Specifically, DenseTrans is a densely connected Transformer, which learns semantic embeddings for texts based on Transformer layers. Meanwhile, DenseTrans promotes low-level features through dense connections to keep the discriminative power of the learned representations. DenseTrans is inspired by DenseNet in computer vision (CV), but poses a new way to use the dense connectivity which is totally different from its original design purpose. Experimental results over two question retrieval benchmark datasets show that our model can obtain significant gain on recall against strong term-based methods as well as state-of-the-art embedding-based methods.
Semantic Hashing is a popular family of methods for efficient similarity search in large-scale datasets. In Semantic Hashing, documents are encoded as short binary vectors (i.e., hash codes), such that semantic similarity can be efficiently computed using the Hamming distance. Recent state-of-the-art approaches have utilized weak supervision to train better performing hashing models. Inspired by this, we present Semantic Hashing with Pairwise Reconstruction (PairRec), which is a discrete variational autoencoder based hashing model. PairRec first encodes weakly supervised training pairs (a query document and a semantically similar document) into two hash codes, and then learns to reconstruct the same query document from both of these hash codes (i.e., pairwise reconstruction). This pairwise reconstruction enables our model to encode local neighbourhood structures within the hash code directly through the decoder. We experimentally compare PairRec to traditional and state-of-the-art approaches, and obtain significant performance improvements in the task of document similarity search.
In this paper we propose a new document classification method, bridging discrepancies (so-called semantic gap) between the training set and the application sets of textual data. We demonstrate its superiority over classical text classification approaches, including traditional classifier ensembles. The method consists in combining a document categorization technique with a single classifier or a classifier ensemble (SEMCOM algorithm - Committee with Semantic Categorizer).
We propose a neural network architecture for learning vector representations of hotels. Unlike previous works, which typically only use user click information for learning item embeddings, we propose a framework that combines several sources of data, including user clicks, hotel attributes (e.g., property type, star rating, average user rating), amenity information (e.g., the hotel has free Wi-Fi or free breakfast), and geographic information. During model training, a joint embedding is learned from all of the above information. We show that including structured attributes about hotels enables us to make better predictions in a downstream task than when we rely exclusively on click data. We train our embedding model on more than 40 million user click sessions from a leading online travel platform and learn embeddings for more than one million hotels. Our final learned embeddings integrate distinct sub-embeddings for user clicks, hotel attributes, and geographic information, providing an interpretable representation that can be used flexibly depending on the application. We show empirically that our model generates high-quality representations that boost the performance of a hotel recommendation system in addition to other applications. An important advantage of the proposed neural model is that it addresses the cold-start problem for hotels with insufficient historical click information by incorporating additional hotel attributes which are available for all hotels.
Much of human knowledge is encoded in text, available in scientific publications, books, and the web. Given the rapid growth of these resources, we need automated methods to extract such knowledge into machine-processable structures, such as knowledge graphs. An important task in this process is entity normalization, which consists of mapping noisy entity mentions in text to canonical entities in well-known reference sets. However, entity normalization is a challenging problem; there often are many textual forms for a canonical entity that may not be captured in the reference set, and entities mentioned in text may include many syntactic variations, or errors. The problem is particularly acute in scientific domains, such as biology. To address this problem, we have developed a general, scalable solution based on a deep Siamese neural network model to embed the semantic information about the entities, as well as their syntactic variations. We use these embeddings for fast mapping of new entities to large reference sets, and empirically show the effectiveness of our framework in challenging bio-entity normalization datasets.