learning to rank

Supervised learningsupervised learning, n.: Teaching a computer by example: you give it many inputs paired with the right answers, such as photos labeled 'cat' or 'dog', and it learns to answer correctly for new inputs. of a scoring function whose job is to put items in the right order, most often the documents a search engine retrieves for a query. Its examples are queries paired with candidate items and graded relevance labels or preferences inferred from clicks, with each query–item pair described by features such as text-match scores (BM25BM25, prop. n.: A decades-old formula that search engines use to score how well a document matches a query, rewarding rare query words and repeated matches — with diminishing returns — and correcting for long documents.), link authority and freshness. Methods are grouped by loss: pointwise methods predict each item's relevance independently; pairwise methods (RankNet, RankSVM) learn which of two items should rank higher; and listwise methods optimize a loss over the whole ranked list, often a smooth proxy for position-weighted metrics such as normalized discounted cumulative gainnormalized discounted cumulative gain, n.: A score for a ranked list of results that rewards putting the most relevant items near the top, scaled so that a perfect ordering scores 1.. LambdaMART — gradient-boostedgradient boosting, n.: A method that builds a strong predictor by adding small decision trees one at a time, each trained to correct the errors that remain after all the trees before it. trees driven by LambdaRank's metric-weighted gradients — became the standard and is built into XGBoostXGBoost, prop. n.: A fast, free software tool that makes predictions by adding up hundreds of small decision trees, each one fixing the mistakes of those before it. For years it was behind many winning entries in competitions on spreadsheet-style data., LightGBMLightGBM, prop. n.: A widely used tool for making predictions from very large tables of data, quickly and in little memory, by combining many small decision trees. Microsoft released it as free software in 2016. and CatBoostCatBoost, prop. n.: A free gradient-boosting library from the tech company Yandex, known for handling categories such as city names or product types directly, without first converting them into numbers by hand.. Neural rerankersreranking, n.: A second, more careful pass over search results: a fast method grabs a few hundred candidates, then a slower, smarter model reorders them so the best come first., such as cross-encoderscross-encoder, n.: An AI model that reads a question and a candidate passage together, side by side, and outputs one relevance score. Accurate but slow, so it is used to reorder a short list., now often produce the final ordering in search and retrieval-augmented generationretrieval-augmented generation, n.: Giving an AI an open-book exam: before answering, it looks up relevant passages in a set of documents and bases its reply on them, often citing where each point came from. pipelines.

We replaced the hand-tuned scoring formula with a LambdaMART learning-to-rank model trained on click logs.