data is suggesting alternative queries when user queries contain typographical errors. Query suggestion thus can improve users’ search experience by avoiding returning empty result or results of poor qualities. In this paper, we study the problem of effectively and efficiently providing quality query suggestions for keyword queries on an XML document. We illustrate certain biases in previous work and propose a principled and general framework, XClean, based on the state-of-the-art language model. Compared with previous methods, XClean can accommodate different error models and XML keyword query semantics without losing rigor. Algorithms have been developed that compute the top-k suggestions efficiently. We performed an extensive experiment study using two large-scale real datasets. The experiment results demonstrate the effectiveness and efficiency of the proposed methods. I.

We present a web-based algorithm for the task of POS tagging of unknown words (words appearing only a small number of times in the training data of a supervised POS tagger). When a sentence s containing an unknown word u is to be tagged by a trained POS tagger, our algorithm collects from the web contexts that are partially similar to the context of u in s, which are then used to compute new tag assignment probabilities for u. Our algorithm enables fast multi-domain unknown word tagging, since, unlike previous work, it does not require a corpus from the new domain. We integrate our algorithm into the MXPOST POS tagger (Ratnaparkhi, 1996) and experiment with three languages (English, German and Chinese) in seven in-domain and domain adaptation scenarios. Our algorithm provides an error reduction of up to 15.63% (English), 18.09 % (German) and 13.57% (Chinese) over the original tagger. 1

Our work concerns the design of robust information retrieval environments that can successfully handle queries containing misspelled words. Our aim is to perform a comparative analysis of the efficacy of two possible strategies that can be adopted. A first strategy involves those approaches based on correcting the misspelled query, thus requiring the integration of linguistic information in the system. This solution has been studied from complementary standpoints, according to whether contextual information of a linguistic nature is integrated in the process or not, the former implying a higher degree of complexity. A second strategy involves the use of character n-grams as the basic indexing unit, which guarantees the robustness of the information retrieval process whilst at the same time eliminating the need for a specific query correction stage. This is a knowledgelight and language-independent solution which requires no linguistic information for its application. Both strategies have been subjected to experimental testing, with Spanish being

Spelling correction for keyword-search queries is challenging in restricted domains such as personal email (or desktop) search, due to the scarcity of query logs, and due to the specialized nature of the domain. For that task, this paper presents an algorithm that is based on statistics from the corpus data (rather than the query log). This algorithm, which employs a simple graph-based approach, can incorporate different types of data sources with different levels of reliability (e.g., email subject vs. email body), and can handle complex spelling errors like splitting and merging of words. An experimental study shows the superiority of the algorithm over existing alternatives in the email domain. 1

Chinese Pinyin input method is very important for Chinese language information processing. Users may make errors when they are typing in Chinese words. In this paper, we are concerned with the reasons that cause the errors. Inspired by the observation that pressing backspace is one of the most common user behaviors to modify the errors, we collect 54, 309, 334 error-correction pairs from a realworld data set that contains 2, 277, 786 users via backspace operations. In addition, we present a comparative analysis of the data to achieve a better understanding of users ’ input behaviors. Comparisons with English typos suggest that some language-specific properties result in a part of Chinese input errors. 1

Query spelling correction is a crucial component of moden search engines that can help users to express an information need more accurately and thus improve search quality. In participation of the Microsoft Speller Challenge, we proposed and implemented an efficient end-to-end speller correction system, namely CloudSpeller. The CloudSpeller system uses a Hidden Markov model to effectively model major types of spelling errors in a unified framework, in which we integrate a large-scale lexicon constructed using Wikipedia, an error model trained from high confidence correction pairs, and the Microsoft Web N-gram service. Our system achieves excellent performance on two search query spelling correction datasets, reaching 0.970 and 0.940 F1 scores on the TREC dataset and the MSN dataset respectively.

The search in patent databases is a risky business compared to the search in other domains. A single document that is relevant but overlooked during a patent search can turn into an expensive proposition. While recent research engages in specialized models and algorithms to improve the effectiveness of patent retrieval, we bring another aspect into focus: the detection and exploitation of patent inconsistencies. In particular, we analyze spelling errors in the assignee field of patents granted by the United States Patent & Trademark Office. We introduce technology in order to improve retrieval effectiveness despite the presence of typographical ambiguities. In this regard, we (1) quantify spelling errors in terms of edit distance and phonological dissimilarity and (2) render error detection as a learning problem that combines word dissimilarities with patent meta-features. For the task of finding all patents of a company, our approach improves recall from 96.7% (when using a state-of-the-art patent search engine) to 99.5%, while precision is compromised by only 3.7%. 1

Discriminative training in query spelling correction is difficult due to the complex internal structures of the data. Recent work on query spelling correction suggests a two stage approach a noisy channel model that is used to retrieve a number of candidate corrections, followed by discriminatively trained ranker applied to these candidates. The ranker, however, suffers from the fact the low recall of the first, suboptimal, search stage. This paper proposes to directly optimize the search stage with a discriminative model based on latent structural SVM. In this model, we treat query spelling correction as a multiclass classification problem with structured input and output. The latent structural information is used to model the alignment of words in the spelling correction process. Experiment results show that as a standalone speller, our model outperforms all the baseline systems. It also attains a higher recall compared with the noisy channel model, and can therefore serve as a better filtering stage when combined with a ranker. 1