Retrieval-Augmented Generation (RAG) is a prevalent approach to infuse a private knowledge base of documents with Large Language Models (LLM) to build Generative Q\&A (Question-Answering) systems. However, RAG accuracy becomes increasingly challenging as the corpus of documents scales up, with Retrievers playing an outsized role in the overall RAG accuracy by extracting the most relevant document from the corpus to provide context to the LLM. In this paper, we propose the 'Blended RAG' method of leveraging semantic search techniques, such as Dense Vector indexes and Sparse Encoder indexes, blended with hybrid query strategies. Our study achieves better retrieval results and sets new benchmarks for IR (Information Retrieval) datasets like NQ and TREC-COVID datasets. We further extend such a 'Blended Retriever' to the RAG system to demonstrate far superior results on Generative Q\&A datasets like SQUAD, even surpassing fine-tuning performance.

Machine Reading Comprehension (MRC) models tend to take advantage of spurious correlations (also known as dataset bias or annotation artifacts in the research community). Consequently, these models may perform the MRC task without fully comprehending the given context and question, which is undesirable since it may result in low robustness against distribution shift. The main focus of this paper is answer-position bias, where a significant percentage of training questions have answers located solely in the first sentence of the context. We propose a Single-Sentence Reader as a new approach for addressing answer position bias in MRC. Remarkably, in our experiments with six different models, our proposed Single-Sentence Readers trained on biased dataset achieve results that nearly match those of models trained on normal dataset, proving their effectiveness in addressing the answer position bias. Our study also discusses several challenges our Single-Sentence Readers encounter and proposes a potential solution.

With a rise in false, inaccurate, and misleading information in propaganda, news, and social media, real-world Question Answering (QA) systems face the challenges of synthesizing and reasoning over contradicting information to derive correct answers. This urgency gives rise to the need to make QA systems robust to misinformation, a topic previously unexplored. We study the risk of misinformation to QA models by investigating the behavior of the QA model under contradicting contexts that are mixed with both real and fake information. QA, which contains over 10K human-written and model-generated contradicting pairs of contexts. Experiments show that QA models are vulnerable under contradicting contexts brought by misinformation. To defend against such threat, we build a misinformation-aware QA system as a counter-measure that integrates question answering and misinformation detection in a joint fashion. A typical Question Answering (QA) system (Chen et al., 2017; Yang et al., 2019; Karpukhin et al., 2020; Lewis et al., 2020b) starts by retrieving a set of relevant context documents from the Web, which are then examined by a machine reader to identify the correct answer. Existing work equate Wikipedia as the web corpus. Therefore, all retrieved context documents are assumed to be clean and trustable. However, real-world QA faces a much noisier environment, where the web corpus is tainted with misinformation.

Current textual question answering models achieve strong performance on in-domain test sets, but often do so by fitting surface-level patterns in the data, so they fail to generalize to out-of-distribution settings. To make a more robust and understandable QA system, we model question answering as an alignment problem. We decompose both the question and context into smaller units based on off-the-shelf semantic representations (here, semantic roles), and align the question to a subgraph of the context in order to find the answer. We formulate our model as a structured SVM, with alignment scores computed via BERT, and we can train end-to-end despite using beam search for approximate inference. Our explicit use of alignments allows us to explore a set of constraints with which we can prohibit certain types of bad model behavior arising in cross-domain settings. Furthermore, by investigating differences in scores across different potential answers, we can seek to understand what particular aspects of the input lead the model to choose the answer without relying on post-hoc explanation techniques. We train our model on SQuAD v1.1 and test it on several adversarial and out-of-domain datasets. The results show that our model is more robust cross-domain than the standard BERT QA model, and constraints derived from alignment scores allow us to effectively trade off coverage and accuracy.

A recent approach to the popular extractive question answering (extractive QA) task that generates its own training data instead of requiring existing annotated question answering examples. Extractive QA is a popular task for natural language processing (NLP) research, where models must extract a short snippet from a document in order to answer a natural language question. Though supervised models perform well at extractive QA, they require thousands -- sometimes hundreds of thousands -- of annotated examples for training, and their performance suffers when tested outside of the textual domains and language they were trained on. By approaching extractive QA as a self-supervised task, our technique outperformed early supervised models on the widely used SQuAD data set while requiring no annotated question answering training data. The code for our method is now available to download.

Applying neural-networks on Question Answering has gained increasing popularity in recent years. In this paper, I implemented a model with Bi-directional attention flow layer, connected with a Multi-layer LSTM encoder, connected with one start-index decoder and one conditioning end-index decoder. I introduce a new end-index decoder layer, conditioning on start-index output. The Experiment shows this has increased model performance by 15.16%. For prediction, I proposed a new smart-span equation, rewarding both short answer length and high probability in start-index and end-index, which further improved the prediction accuracy. The best single model achieves an F1 score of 73.97% and EM score of 64.95% on test set.

Two artificial intelligence programs created by Chinese e-commerce company Alibaba and Microsoft beat humans on a Stanford University reading comprehension test. Alibaba took the honor as the creator of the first program to ever beat a human in a reading comprehension test, scoring 82.44 percent and narrowly edging past the human's 82.304 percent. A different program built by Microsoft scored higher than Alibaba's at 82.605 percent. Microsoft's took the same test as Alibaba's but was finalized a day later, according to Bloomberg. The test known as Stanford Question Answering Dataset, or SQuAD for short, asks the contestants – human and robot – to provide exact answers to more than 100,000 questions drawn from more than 500 Wikipedia articles.

At least 80% of enterprise data is unstructured, contained in the myriad text-based social conversations that are happening every day. Unlocking the hidden value of text through predictive analytics is imperative to the understanding of customers' opinions and needs, to make better, more informed business decisions. A whopping 90% of this data is actually completely underutilized when it comes to data strategies and data analytics techniques. It's very easy for humans to consume and make sense of unstructured data, but machines don't find it as easy. At the rate it's being created, it's almost impossible for humans to consume this information at the rate that it's growing.

In this teaser for an upcoming Super Bowl ad, Baldwin gets help planning a Super Bowl party from Marino and the Amazon Echo. SAN FRANCISCO – After 22 years of sitting the game out, Amazon decided to make its first Super Bowl ad about Echo -- a decision that underlined how aggressively Amazon is pushing the personal assistant/wireless speaker to the mass market. The Seattle online retailer launched Echo, a voice-driven, cloud-connected, wireless speaker last June. It ranked as a favorite entry among many tech reviewers in the new and growing Internet of Things market. But while the Echo has "a cult following, it needs more consumer awareness," said Sucharita Mulpuru, Amazon analyst for Forrester Research.