--- title: "Shankar Kumar — author profile and papers" description: "Shankar Kumar, Google (United States): 2 papers on Elamur Research (2004), h-index 26 — affiliations, topics, co-authors and who cites them in our library." canonical: "https://research.elamur.ai/free/en/authors/shankar-kumar/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/authors/shankar-kumar/) [home](https://md.research.elamur.ai/free/en/) / [authors](https://md.research.elamur.ai/free/en/authors/) / Shankar Kumar # Shankar Kumar Shankar Kumar works on statistical [machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/), and in particular on how a translation system should choose among the candidate outputs it produces. A recurring question in the papers here is how linguistic knowledge, especially syntax, can shape that choice without redesigning the underlying decoder. The methods centre on lists of candidate translations, evaluation metrics such as BLEU, and parse trees of source and target sentences. The most cited work on this site is [Minimum Bayes-Risk Decoding for Statistical Machine Translation](https://md.research.elamur.ai/free/en/papers/kumar-2004-minimum-bayes-risk-decoding-statistical/) (2004), written with Byrne. It takes minimum Bayes-risk decoding, a technique earlier applied to speech recognition and to word alignment of parallel text, and carries it over to translation. Rather than returning the single most probable output, the decoder picks the candidate whose expected loss is lowest under a loss function chosen by the user. The paper groups such losses into lexical, target-side parse-tree and bilingual parse-tree levels, and proposes a Bitree loss that compares parse structures across the two languages. The second, [A Smorgasbord of Features for Statistical Machine Translation](https://md.research.elamur.ai/free/en/papers/och-2004-smorgasbord-features-statistical-machine-translation/) (2004), is a team study of Chinese-English translation. It leaves a phrase-based alignment template system unchanged and instead reranks long n-best lists with a log-linear model. The added features range from words and part-of-speech tags to chunks and full parses, with weights tuned directly for BLEU. Both papers here date from 2004, so no later shift in direction can be traced on this site. Together they show two related routes for adding syntax to translation: building it into the loss used at decoding time, or using it in features for reranking. Kumar is currently at Google in Mountain View, with bylines there running from 2006 to 2025, and earlier papers from Johns Hopkins University. Co-authors on the feature-reranking study include [Franz Josef Och](https://md.research.elamur.ai/free/en/authors/franz-josef-och/), [Daniel Gildea](https://md.research.elamur.ai/free/en/authors/daniel-gildea/), [Sanjeev Khudanpur](https://md.research.elamur.ai/free/en/authors/sanjeev-khudanpur/) and [Alexander Fraser](https://md.research.elamur.ai/free/en/authors/alexander-fraser/). 2 papers on this site · 2004 current affiliation [Google (United States)](https://md.research.elamur.ai/free/en/companies/google/), Mountain View, United States Shankar Kumar has 2 papers on this site, in 2004, most of them on [Machine translation](https://md.research.elamur.ai/free/en/topics/machine-translation/); often with [Alexander Fraser](https://md.research.elamur.ai/free/en/authors/alexander-fraser/), [Daniel Gildea](https://md.research.elamur.ai/free/en/authors/daniel-gildea/), [David Smith](https://md.research.elamur.ai/free/en/authors/david-smith/). [OpenAlex](https://openalex.org/A5101051347) · [ORCID](https://orcid.org/0009-0001-4307-8102) · [**h-index 26**](https://openalex.org/authors/A5101051347) · [76 works](https://openalex.org/works?filter=author.id:A5101051347) · [3 476 citations](https://openalex.org/works?filter=author.id:A5101051347&sort=cited_by_count:desc) OpenAlex, 2026-10-14 [2 works on this site](https://md.research.elamur.ai/free/en/authors/shankar-kumar/#papers) · [607 citations in our library](https://md.research.elamur.ai/free/en/authors/shankar-kumar/cited-by/) · 4 more works in our library The profile is the library's match of the name; two people with one name are not always told apart. Elsewhere: [Google Scholar](https://scholar.google.com/citations?view_op=search_authors&mauthors=Shankar+Kumar) · [Semantic Scholar](https://www.semanticscholar.org/search?q=Shankar+Kumar) · [dblp](https://dblp.org/search?q=Shankar+Kumar) · [ACL Anthology](https://aclanthology.org/search/?q=Shankar+Kumar) ## Affiliations - [Google (United States)](https://md.research.elamur.ai/free/en/companies/google/) · Mountain View, United States —current · 2006–2025 on 45 papers - [Johns Hopkins University](https://md.research.elamur.ai/free/en/universities/johns-hopkins-university/) · Baltimore, United States —2001–2023 on 11 papers From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 1 more institution on single bylines are not listed. ## Topics [Machine translation 2](https://md.research.elamur.ai/free/en/topics/machine-translation/) ## Works by year ### 2000s [**2004** 2 papers](https://md.research.elamur.ai/free/en/authors/shankar-kumar/year/2004/) ## Papers The most cited first (OpenAlex and our library). 1. **[Minimum Bayes-Risk Decoding for Statistical Machine Translation](https://md.research.elamur.ai/free/en/papers/kumar-2004-minimum-bayes-risk-decoding-statistical/)** *(2004 · 583 citations)* Statistical MT systems usually pick the most probable translation (MAP decoding), even though they are judged by metrics such as BLEU, WER or PER that capture different aspects of quality. 2. **[A Smorgasbord of Features for Statistical Machine Translation](https://md.research.elamur.ai/free/en/papers/och-2004-smorgasbord-features-statistical-machine-translation/)** *(2004 · 402 citations)* The paper asks whether richer linguistic knowledge, especially syntax, can fix typical errors of statistical Chinese-English MT. ## Also in our library 4 more works of theirs are in our library without a page here yet; they will appear as we publish them. ## Co-authors here 7 co-authors with a page, the most papers together first. - [Alexander Fraser](https://md.research.elamur.ai/free/en/authors/alexander-fraser/) - [Daniel Gildea](https://md.research.elamur.ai/free/en/authors/daniel-gildea/) - [David Smith](https://md.research.elamur.ai/free/en/authors/david-smith/) - [Franz Josef Och](https://md.research.elamur.ai/free/en/authors/franz-josef-och/) - [Kenji Yamada](https://md.research.elamur.ai/free/en/authors/kenji-yamada/) - [Libin Shen](https://md.research.elamur.ai/free/en/authors/libin-shen/) - [Sanjeev Khudanpur](https://md.research.elamur.ai/free/en/authors/sanjeev-khudanpur/) ## Cited by Papers of our library cite their works 607 times (of the 140 884 papers whose bibliographies we read); [35 citing papers with a page here →](https://md.research.elamur.ai/free/en/authors/shankar-kumar/cited-by/).