We conducted a manual error analysis and found that our QA model typically performed well on questions involving knowledge, commonsense reasoning, and definitions, but it often struggled to understand wordplay or theme-related clues.Īfter running the QA model on each clue, the BCS runs loopy belief propagation to iteratively update the answer probabilities in the grid. Compared to the previous state-of-the-art method for answering crossword clues, this approach obtained a 13.4% absolute improvement in top-1000 QA accuracy. Rather than passages, however, our approach maps both questions and answers into a shared embedding space and finds answers directly. The BCS’s question answering model is based on DPR (Karpukhin et al., 2020), which is a bi-encoder model typically used to retrieve passages that are relevant to a given question. First, it generates a probability distribution over possible answers to each clue using a question answering (QA) model second, it uses probabilistic inference, combined with local search and a generative language model, to handle conflicts between proposed intersecting answers.įigure 2: Architecture diagram of the Berkeley Crossword Solver The BCS uses a two-step process to solve crossword puzzles. Domain ender that UC Berkeley was one of the first schools to adopt (3).They’re given out at Berkeley’s HAAS School (4).Here are a handful of example clues from our dataset (answers at the bottom of this post): While some clues are similar to factoid question answering, others require relational reasoning or understanding difficult wordplay. Many clues are vague or underspecified and can’t be answered until crossing constraints are taken into account.
TIMES CROSSWORD EDITOR WILL CROSSWORD CODE
The most recent version is the current top-performing system on crossword puzzles from The New York Times, achieving 99.7% letter accuracy (see the technical paper, web demo, and code release).Ĭrosswords are challenging for humans and computers alike. The BCS combines neural question answering and probabilistic inference to achieve near-perfect performance on most American-style crossword puzzles, like the one shown below:įigure 1: Example American-style crossword puzzleĪn earlier version of the BCS, in conjunction with Dr.Fill, was the first computer program to outscore all human competitors in the world’s top crossword tournament. We recently published the Berkeley Crossword Solver (BCS), the current state of the art for solving American-style crossword puzzles. Eric Wallace, Nicholas Tomlin, Albert Xu, Kevin Yang, Eshaan Pathak
0 kommentar(er)
