Citation Context

... parser. We have two methods for deriving base-phrase annotations for a string. First, we trained a finitestate shallow parser on base phrases extracted from the Penn Wall St. Journal (WSJ) Treebank (=-=Marcus et al., 1993-=-). The treebank trees are pre-processed identically to the procedure for training the Charniak parser, e.g., empty nodes and function tags are removed. The shallow parser is trained using the perceptr...

Citation Context

...st solutions may be pruned in early stages. Pipeline systems are ubiquitous in natural language processing, used not only in parsing (Ratnaparkhi, 1999; Charniak, 2000), but also machine translation (=-=Och and Ney, 2003-=-) and speech recognition (Fiscus, 1997; Goel et al., 2000), among others. Despite the widespread use of pipelines, they have been understudied, with very little work on general techniques for designin...

Citation Context

...n later stages, usually at the risk that the best solutions may be pruned in early stages. Pipeline systems are ubiquitous in natural language processing, used not only in parsing (Ratnaparkhi, 1999; =-=Charniak, 2000-=-), but also machine translation (Och and Ney, 2003) and speech recognition (Fiscus, 1997; Goel et al., 2000), among others. Despite the widespread use of pipelines, they have been understudied, with v...

Citation Context

...ed to find solutions within that subset. For example, a part-of-speech tagger could constrain a “base phrase” chunker (Ratnaparkhi, 1999), or the n-best output of a parser could constrain a reranker (=-=Charniak and Johnson, 2005-=-). A pipeline is typically used to reduce search complexity for rich models used in later stages, usually at the risk that the best solutions may be pruned in early stages. Pipeline systems are ubiqui...

Citation Context

...ximum sum of the posterior probabilities of its constituents. This is done by computing the posterior probabilities of constituents in a chart, typically via the InsideOutside algorithm (Baker, 1979; =-=Lari and Young, 1990-=-), followed by a final CYK-like pass to find the tree maximizing the sum. For non-binary branching trees, where precision and recall may differ, Goodman (1998, Ch.3) proposes the following combined me...

Citation Context

... Pipeline systems are ubiquitous in natural language processing, used not only in parsing (Ratnaparkhi, 1999; Charniak, 2000), but also machine translation (Och and Ney, 2003) and speech recognition (=-=Fiscus, 1997-=-; Goel et al., 2000), among others. Despite the widespread use of pipelines, they have been understudied, with very little work on general techniques for designing and improving pipeline systems (alth...

Citation Context

...e with the maximum sum of the posterior probabilities of its constituents. This is done by computing the posterior probabilities of constituents in a chart, typically via the InsideOutside algorithm (=-=Baker, 1979-=-; Lari and Young, 1990), followed by a final CYK-like pass to find the tree maximizing the sum. For non-binary branching trees, where precision and recall may differ, Goodman (1998, Ch.3) proposes the...

Citation Context

...ine identifies a subset of the possible solutions, and later stages are constrained to find solutions within that subset. For example, a part-of-speech tagger could constrain a “base phrase” chunker (=-=Ratnaparkhi, 1999-=-), or the n-best output of a parser could constrain a reranker (Charniak and Johnson, 2005). A pipeline is typically used to reduce search complexity for rich models used in later stages, usually at t...

Citation Context

.... 2.1 Base Phrases Following Ratnaparkhi (1999), we define a base phrase as any parse node with only preterminal children. Unlike the shallow phrases defined for the CoNLL-2000 Shared Task (Tjong Kim =-=Sang and Buchholz, 2000-=-), base phrases correspond directly to constituents that appear in full parses, and hence can provide a straightforward constraint on edges within a chart parser. In contrast, shallow phrases collapse...

Citation Context

... and Charniak, 1998; Blaheta and Charniak, 1999; Hall and Johnson, 2004; Charniak et al., 2006) as well as many focused on optimizing final parse accuracy (Charniak, 2000; Charniak and Johnson, 2005; =-=McClosky et al., 2006-=-). This focus on optimization has made system improvements very difficult to achieve; yet our relatively simple architecture yields statistically significant improvements, making pipeline iteration a ...

Citation Context

...and reranker output. The Charniak parsing pipeline has been extensively studied over the past decade, with a number of papers focused on improving early stages of the pipeline (Charniak et al., 1998; =-=Caraballo and Charniak, 1998-=-; Blaheta and Charniak, 1999; Hall and Johnson, 2004; Charniak et al., 2006) as well as many focused on optimizing final parse accuracy (Charniak, 2000; Charniak and Johnson, 2005; McClosky et al., 20...

Citation Context

... shallow parser output and reranker output. The Charniak parsing pipeline has been extensively studied over the past decade, with a number of papers focused on improving early stages of the pipeline (=-=Charniak et al., 1998-=-; Caraballo and Charniak, 1998; Blaheta and Charniak, 1999; Hall and Johnson, 2004; Charniak et al., 2006) as well as many focused on optimizing final parse accuracy (Charniak, 2000; Charniak and John...

Citation Context

...y constraining parses to be consistent with a base-phrase tree. We derive these base-phrase constraints from three sources: the reranking stage of the parsing pipeline; a finite-state shallow parser (=-=Hollingshead et al., 2005-=-); and a combination of the output from these two sources. We compare the relative performance of these three sources and find the best performance improvements using constraints derived from a weight...

Citation Context

...iak parsing pipeline has been extensively studied over the past decade, with a number of papers focused on improving early stages of the pipeline (Charniak et al., 1998; Caraballo and Charniak, 1998; =-=Blaheta and Charniak, 1999-=-; Hall and Johnson, 2004; Charniak et al., 2006) as well as many focused on optimizing final parse accuracy (Charniak, 2000; Charniak and Johnson, 2005; McClosky et al., 2006). This focus on optimizat...

Citation Context

...ems are ubiquitous in natural language processing, used not only in parsing (Ratnaparkhi, 1999; Charniak, 2000), but also machine translation (Och and Ney, 2003) and speech recognition (Fiscus, 1997; =-=Goel et al., 2000-=-), among others. Despite the widespread use of pipelines, they have been understudied, with very little work on general techniques for designing and improving pipeline systems (although cf. Finkel et ...

Citation Context

...er the past decade, with a number of papers focused on improving early stages of the pipeline (Charniak et al., 1998; Caraballo and Charniak, 1998; Blaheta and Charniak, 1999; Hall and Johnson, 2004; =-=Charniak et al., 2006-=-) as well as many focused on optimizing final parse accuracy (Charniak, 2000; Charniak and Johnson, 2005; McClosky et al., 2006). This focus on optimization has made system improvements very difficult...