Showing 1-8 of 304 results

Geographic microtargeting of social assistance with high-resolution poverty maps

Publication Date: 2022
Creators: Smythe, Isabella S.; Blumenstock Joshua E.

Hundreds of millions of poor families receive some form of targeted social assistance. Many of these antipoverty programs involve some degree of geographic targeting, where aid is prioritized to the poorest regions of the country. However, policy makers in many low-resource settings lack the disaggregated poverty data required to make effective geographic targeting decisions. Using several independent datasets from Nigeria, this paper shows that high-resolution poverty maps, constructed by applying machine learning algorithms to satellite imagery and other nontraditional geospatial data, can improve the targeting of government cash transfers to poor families. Specifically, we find that geographic targeting relying on machine learning–based poverty maps can reduce errors of exclusion and inclusion relative to geographic targeting based on recent nationally representative survey data. This result holds for antipoverty programs that target both the poor and the extreme poor and for initiatives of varying sizes. We also find no evidence that machine learning–based maps increase targeting disparities by demographic groups, such as gender or religion. Based in part on these findings, the Government of Nigeria used this approach to geographically target emergency cash transfers in response to the COVID-19 pandemic.

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Publication Date: 2019
Creators: Devlin, Jacob; Chang, Ming-Wie; Lee, Kenton; Toutanova, Kristina

We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models, BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications.
BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5% (7.7% point absolute improvement), MultiNLI accuracy to 86.7% (4.6% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).

Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing

Publication Date: 2023
Creators: Pengfei, Liu; Yuan, Weizhe; Fu, Jinlan; Jiang, Zhengbao; Hayashi, Hiroaki; Neubig, Graham

This article surveys and organizes research works in a new paradigm in natural language processing, which we dub “prompt-based learning.” Unlike traditional supervised learning, which trains a model to take in an input x and predict an output y as P(y|x), prompt-based learning is based on language models that model the probability of text directly. To use these models to perform prediction tasks, the original input x is modified using a template into a textual string prompt x′ that has some unfilled slots, and then the language model is used to probabilistically fill the unfilled information to obtain a final string x̂, from which the final output y can be derived. This framework is powerful and attractive for a number of reasons: It allows the language model to be pre-trained on massive amounts of raw text, and by defining a new prompting function the model is able to perform few-shot or even zero-shot learning, adapting to new scenarios with few or no labeled data. In this article, we introduce the basics of this promising paradigm, describe a unified set of mathematical notations that can cover a wide variety of existing work, and organize existing work along several dimensions, e.g., the choice of pre-trained language models, prompts, and tuning strategies. To make the field more accessible to interested beginners, we not only make a systematic review of existing works and a highly structured typology of prompt-based concepts but also release other resources, e.g., a website NLPedia–Pretrain including constantly updated survey and paperlist.

Sentiment Analysis of Movie Review using Hybrid Optimization with Convolutional Neural Network in English Language

Publication Date: 2023
Creators: Tidake, Vishal & Mazumdar, Nilanjan & Kumar, A. & Rao, B. & Fatma, Dr Gulnaz & Raj, I.. (2023). Sentiment Analysis of Movie Review using Hybrid Optimization with Convolutional Neural Network in English Language. 1668-1673. 10.1109/ICAIS56108.2023.10073750.

There seems to be a growing amount of user-generated material online as more people become familiar with the Internet. Understanding hidden thoughts, emotions, and attitudes in tweets, emails, comments, and reviews is difficult yet essential for market analysis, brand tracking, social media tracking, and customer support. Sentiment Analysis (SA) identifies the emotional undertone of a string of words and also might basically be employed to comprehend a user’s attitude, thoughts, and emotions. The Harris Hawks Optimization – Sparrow S earch Algorithm with Convolutional Neural Network i.e., (HH-SSA-CNN) proposed in this study is an innovative SA algorithm. Pre-processing, sentiment categorization, and feature extraction make up the procedure. The preprocessing phase removes the unwanted info from input text evaluations using NLP algorithms. A hybrid technique that combines review-related features and aspect-related features has been presented for efficiently retrieving the features. This method creates unique composite features for every review. The created HH- SSA-CNN is used to accomplish sentiment categorization. This approach has been used in the IMDb dataset. To assess the model’s efficacy, the outcomes of the HH-SSA-CNN model are contrasted with those of alternative methodologies. The result indicates that the developed model accurately classifies the sentiments while compared to other existing methods.

The traditional recommendation system can hardly utilize the ingredients and flavor characteristics of the dishes, and faces the problems of sparse data and cold start, resulting in inaccurate results of the recommendation system. This paper addresses these problems by constructing a heterogeneous graph network with the interaction data between users and dishes in the food domain and taking the main and auxiliary ingredients as nodes. Also, we capture the higher-order structural information among users, dishes, and main and auxiliary ingredients by using meta-path guidance. In the meantime, we assign weights to the edges associated with user nodes and main and auxiliary ingredients nodes, which can obtain users’ preferences for main and auxiliary ingredients by using weighted GCN networks. The experiments conducted on a food domain dataset demonstrate that the meta-path-guided heterogeneous graph dish recommendation algorithm proposed in this paper is improved over the traditional recommendation algorithm.

Measuring the Impact of Explanation Bias: A Study of Natural Language Justifications for Recommender Systems

Publication Date: 2023
Creators: Balog, Krisztian; Radlinski, Filip; Petrov, Andrey

Despite the potential impact of explanations on decision making, there is a lack of research on quantifying their effect on users’ choices. This paper presents an experimental protocol for measuring the degree to which positively or negatively biased explanations can lead to users choosing suboptimal recommendations. Key elements of this protocol include a preference elicitation stage to allow for personalizing recommendations, manual identification and extraction of item aspects from reviews, and a controlled method for introducing bias through the combination of both positive and negative aspects. We study explanations in two different textual formats: as a list of item aspects and as fluent natural language text. Through a user study with 129 participants, we demonstrate that explanations can significantly affect users’ selections and that these findings generalize across explanation formats

A Unifying and General Account of Fairness Measurement in Recommender Systems

Publication Date: 2022
Creators: Amigó, Enrique; Deldjoo, Yashar; Mizzaro, Stefan; Bellogín, Alejandro

Fairness is fundamental to all information access systems, including recommender systems. However, the landscape of fairness definition and measurement is quite scattered with many competing definitions that are partial and often incompatible. There is much work focusing on specific-and different-notions of fairness and there exist dozens of metrics of fairness in the literature, many of them redundant and most of them incompatible. In contrast, to our knowledge, there is no formal framework that covers all possible variants of fairness and allows developers to choose the most appropriate variant depending on the particular scenario. In this paper, we aim to define a general, flexible, and parameterizable framework that covers a whole range of fairness evaluation possibilities. Instead of modeling the metrics based on an abstract definition of fairness, the distinctive feature of this study compared to the current state of the art is that we start from the metrics applied in the literature to obtain a unified model by generalization. The framework is grounded on a general work hypothesis: interpreting the space of users and items as a probabilistic sample space, two fundamental measures in information theory (Kullback-Leibler Divergence and Mutual Information) can capture the majority of possible scenarios for measuring fairness on recommender system outputs. In addition, earlier research on fairness in recommender systems could be viewed as single-sided, trying to optimize some form of equity across either user groups or provider/procurer groups, without considering the user/item space in conjunction, thereby overlooking/disregarding the interplay between user and item groups. Instead, our framework includes the notion of statistical independence between user and item groups. We finally validate our approach experimentally on both synthetic and real data according to a wide range of state-of-the-art recommendation algorithms and real-world data sets, showing that with our framework we can measure fairness in a general, uniform, and meaningful way.

Towards Out-of-Distribution Sequential Event Prediction: A Causal Treatment.

Publication Date: 2022
Creators: Yang, Chenxiao; Wu, Qitian; Wen, Qingsong; Zhou, Zhiqiang; Sun, Liang; Yan, Junchi

The goal of sequential event prediction is to estimate the next event based on a sequence of historical events, with applications to sequential recommendation, user behavior analysis and clinical treatment. In practice, the next-event prediction models are trained with sequential data collected at one time and need to generalize to newly arrived sequences in remote future, which requires models to handle temporal distribution shift from training to testing. In this paper, we first take a data-generating perspective to reveal a negative result that existing approaches with maximum likelihood estimation would fail for distribution shift due to the latent context confounder, i.e., the common cause for the historical events and the next event. Then we devise a new learning objective based on backdoor adjustment and further harness variational inference to make it tractable for sequence learning problems. On top of that, we propose a framework with hierarchical branching structures for learning context-specific representations. Comprehensive experiments on diverse tasks (e.g., sequential recommendation) demonstrate the effectiveness, applicability and scalability of our method with various off-the-shelf models as backbones.

Sign In


Reset Password

Please enter your username or email address, you will receive a link to create a new password via email.