Relation classification, a crucial component of relation extraction, involves identifying connections between two entities. Previous studies have predominantly focused on integrating the attention mechanism into relation classification at a global scale, overlooking the importance of the local context. To address this gap, this paper introduces a novel global-local attention mechanism for relation classification, which enhances global attention with a localized focus. Additionally, we propose innovative hard and soft localization mechanisms to identify potential keywords for local attention. By incorporating both hard and soft localization strategies, our approach offers a more nuanced and comprehensive understanding of the contextual cues that contribute to effective relation classification. Our experimental results on the SemEval-2010 Task 8 dataset highlight the superior performance of our method compared to previous attention-based approaches in relation classification.

We study how well large language models (LLMs) explain their generations with rationales -- a set of tokens extracted from the input texts that reflect the decision process of LLMs. We examine LLM rationales extracted with two methods: 1) attribution-based methods that use attention or gradients to locate important tokens, and 2) prompting-based methods that guide LLMs to extract rationales using prompts. Through extensive experiments, we show that prompting-based rationales align better with human-annotated rationales than attribution-based rationales, and demonstrate reasonable alignment with humans even when model performance is poor. We additionally find that the faithfulness limitations of prompting-based methods, which are identified in previous work, may be linked to their collapsed predictions. By fine-tuning these models on the corresponding datasets, both prompting and attribution methods demonstrate improved faithfulness. Our study sheds light on more rigorous and fair evaluations of LLM rationales, especially for prompting-based ones.

Moral judgement is a complex human reaction that engages cognitive and emotional dimensions. While some of the morality neural correlates are known, it is currently unclear if we can detect moral violation at a single-trial level. In a pilot study, here we explore the feasibility of moral judgement decoding from text stimuli with passive brain-computer interfaces. For effective moral judgement elicitation, we use video-audio affective priming prior to text stimuli presentation and attribute the text to moral agents. Our results show that further efforts are necessary to achieve reliable classification between moral congruency vs. incongruency states. We obtain good accuracy results for neutral vs. morally-charged trials. With this research, we try to pave the way towards neuroadaptive human-computer interaction and more human-compatible large language models (LLMs)

AI models (including LLM) often rely on narrative question-answering (QA) datasets to provide customized QA functionalities to support downstream children education applications; however, existing datasets only include QA pairs that are grounded within the given storybook content, but children can learn more when teachers refer the storybook content to real-world knowledge (e.g., commonsense knowledge). We introduce the FairytaleCQA dataset, which is annotated by children education experts, to supplement 278 storybook narratives with educationally appropriate commonsense knowledge. The dataset has 5,868 QA pairs that not only originate from the storybook narrative but also contain the commonsense knowledge grounded by an external knowledge graph (i.e., ConceptNet). A follow-up experiment shows that a smaller model (T5-large) fine-tuned with FairytaleCQA reliably outperforms much larger prompt-engineered LLM (e.g., GPT-4) in this new QA-pair generation task (QAG). This result suggests that: 1) our dataset brings novel challenges to existing LLMs, and 2) human experts' data annotation are still critical as they have much nuanced knowledge that LLMs do not know in the children educational domain.

Your digitally enabled workforce is swimming in communication at every turn. From email, Slack, Jira and back, there is no end to the volume of text and data corporate employees are tasked with reading. Is it any surprise that our reading habits have changed dramatically over the last decade? It turns out the modern workforce tends to skim and skip rather than read deeply for comprehension and introspection. Many digital influences, from our habitual scrolling through social media to the frantic pace of all-day meetings, have trained employees to minimize diligent reading and thorough comprehension practices.

With the rise in consumption of short format videos and highly personalized content, have you ever thought of having a customized news video feed that would work based on your preferences? This kind of video feed would help us in avoiding redundant news and irrelevant content that we often consume from multiple sources. In this blog, let's make an attempt to automate the process of video creation. In the process, we will be using off the shelf pre-trained models (Fine-tuning these models would increase the performance though) for the sake of simplicity. First, let's select a text source on which we want to make a video on.

Understanding the intent behind email/chat between customers and customer service agents has become a crucial problem nowadays due to an exponential increase in the use of the Internet by people from different cultures and educational backgrounds. More importantly, the explosion of e-commerce has led to a significant increase in text conversation between customers and agents. In this paper, we propose an approach to data mining the conversation intents behind the textual data. Using the customer service dataset, we train unsupervised text representation models using continuous bag of words (CBOW) and Skip-Ngram, and then develop an intent mapping model which would rank the pre-defined intents base on cosine similarity between sentences' embeddings and intents' embeddings. Topic-modeling techniques are used to define intents and domain experts are also involved to interpret topic modelling results. With this approach, we can get a good understanding of the user intentions behind the unlabelled customer service textual data. NTRODUCTION Great amount of customer interactions such as call summaries, email requests, and meeting notes are generated daily by customer service agents.

Artificial Intelligence is the technologies which bestowed us with a fairytale-like performance within our offices and homes. In that picture, the wonder was at the monster's castle, and there wasn't any other living creature there. Here are six strategies to increase your company with artificial intelligence. Science did not create the magic happen in the Beast's Castle -- but science has generated some thing near it to us. The superb technology which may make our lives simpler is called Artificial Intelligence -- and you're able to speak to it, also.

Did you know that many interviews, especially those in the private sector, are being conducted by robots (officially, artificial intelligence...Read More

Early one morning at the beginning of December, I rushed to the nearest newsstand to purchase a copy of the Nikkei Marketing Journal. It was somehow reassuring once again to see, emblazoned atop page one, the publication's traditional sumo-style banzuke (ranking sheet) -- a layout virtually unchanged since 1971 -- listing Japan's top-selling hit products of 2020. Before dissecting the 2020 rankings, it's worth examining how this annual list began. The 1970s marked the time when discerning consumers in Japan began showing a preference for greater variety. Prior to that time, manufacturers had been content to sell their rice cookers, washing machines, TV sets and other mass-produced household items by appealing mainly through brand affinity.