--- title: "JU-SKNSB — paper by Naskar and Bandyopadhyay (2007)" description: "Lesk compares one gloss against another. 8 references, cited by 1 paper here. Among them: 3." authors: ["Sudip Kumar Naskar", "Sivaji Bandyopadhyay"] year: 2007 venue: "Proceedings of the Fourth International Workshop on Semantic Evaluations (SemEval-2007)" source: "https://aclanthology.org/S07-1043/" licence: "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)" licence_url: "https://creativecommons.org/licenses/by-nc-sa/3.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/naskar-2007-ju-sknsb-extended-wordnet-based/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/naskar-2007-ju-sknsb-extended-wordnet-based/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2007](https://md.research.elamur.ai/free/en/papers/2007/page/2/) / JU-SKNSB: Extended WordNet Based WSD on the English All-Words Task at SemEval-1 # JU-SKNSB: Extended WordNet Based WSD on the English All-Words Task at SemEval-1 **Authors:** Sudip Kumar Naskar and [Sivaji Bandyopadhyay](https://md.research.elamur.ai/free/en/authors/sivaji-bandyopadhyay/) · **Year:** 2007 · **Venue:** [Proceedings of the Fourth International Workshop on Semantic Evaluations (SemEval-2007)](https://md.research.elamur.ai/free/en/venues/international-workshop-on-semantic-evaluations/) ([International Workshop on Semantic Evaluations 2007](https://md.research.elamur.ai/free/en/venues/international-workshop-on-semantic-evaluations/2007/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY-NC-SA 3.0 (ACL Anthology, before 2016)](https://creativecommons.org/licenses/by-nc-sa/3.0/) **Links:** [Publisher's page](https://aclanthology.org/S07-1043/) · [PDF](https://aclanthology.org/S07-1043.pdf) · [OpenAlex](https://openalex.org/W2031854707) **Citations:** 3 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/naskar-2007-ju-sknsb-extended-wordnet-based/cited-by/) in our library, of the 149,439 papers whose bibliographies we read ## Summary The paper addresses English all-words word sense disambiguation (WSD) for nouns, verbs and adjectives using only lexical knowledge, with no training data. The authors substantially rework the Lesk overlap method. Lesk compares one gloss against another. Here, each context word is instead matched against "gloss bags" built for every sense of every other word in the window. A gloss bag collects synset members and gloss words from the Extended WordNet, whose glosses carry POS and sense tags. It also includes material from related synsets, chosen per part of speech. A match counts only when POS also agrees. It rewards both sides, with the reward scaled by bag size. All words in the window are resolved jointly, and the window was widened from one sentence to three. Four weighting schemes based on repeat counts were compared. On SemEval-2007 English All-Words (465 instances), the system reached .402 precision and recall, falling back to the WordNet first sense when it had no evidence. On ten Semcor 2.0 files, the best variant reached an F-score of .574. The work shows both the reach and the limits of gloss-overlap WSD built on tagged glosses. ## Key points - **Resources:** Extended WordNet 2.0-1.1, which is built on WordNet 2.0. Predicted senses are converted to WordNet 2.1 using sensemap 2.0 to 2.1, and the authors note this conversion may lose information. - **Relation selection:** a WordNet relation is used for a POS if it accounts for at least 5% of that POS's links. - Nouns: hypernym, hyponym, holonym, meronym, attribute. - Verbs: hyponym, troponym, also see. - Adjectives: attribute, also see, similar to, pertainym. - Exceptions: attribute is kept despite having few links, and antonymy is deliberately left out. - **Semcor results:** evaluated on the first 10 Semcor 2.0 files (9642 instances) with a 3-sentence context. Model C was best (P .621, R .533, F .574); it weights the context word's vote by occurrence count (noc/gbs) and gives the gloss-side sense 1/gbs. The other models scored F .571 (A), .556 (B) and .558 (D). - **Per-POS results for model C:** nouns reached F .6554 and verbs only .4103. The authors attribute the gap partly to noun glosses being the longest in WordNet. They also point to many Extended WordNet tags being of only "silver" or "normal" quality. - **Decision rules:** a word with zero evidence gets no sense from the algorithm, even if it is monosemous. Ties go to the more frequent WordNet sense. Matching cost grows roughly as N²·S·G (words, senses per word, gloss words per bag). ## Interesting ideas - When a context word appears in another word's sense bag, credit flows both ways, to the matched sense and to the bag owner's sense, so every word's sense is chosen as part of one joint assignment. - Each match is divided by the size of the evidence bag, so senses with long definitions do not win just by offering more words to hit. - Relations are admitted by a simple frequency threshold, with reasoned exceptions such as keeping the rare attribute link because it connects sparse adjective data to rich noun data. - Matching uses lemma plus POS plus sense tag rather than raw strings, so only typed, disambiguated gloss tokens count as evidence. - The algorithm abstains when it has no evidence, and the first-sense default is a separate layer, which makes it possible to see how much of the score comes from the method itself. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (world): when anchoring the entities and events of a text to WordNet concepts, mutual-support scoring could pick a consistent set of senses for the whole passage rather than one word at a time. - **I2** (world): any evidence-counting over ontology or knowledge-graph neighbourhoods should normalize by neighbourhood size, so that densely linked concepts do not dominate. - **I3** (world): when compiling the WordNet ontology layer, choose relation types per POS by measured coverage, and document deliberate bridges like adjective→noun attributes. - **I4** (tokens): lemma+POS keys from crate en can serve as the matching unit for lexical lookups, which avoids bag-of-words string coincidences. - **I5** (learn): report method-only results alongside the frequency baseline as a negative control, and tag each decision as evidence-based or defaulted so it stays explainable. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* ## Institutions - [Jadavpur University](https://md.research.elamur.ai/free/en/universities/jadavpur-university/) — University, Kolkata, India · 8 works · 336 citations 📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works 🧭 Topics (2): [English](https://md.research.elamur.ai/free/en/topics/english/) · 2,263 works · [🔀 Word sense disambiguation](https://md.research.elamur.ai/free/en/topics/word-sense-disambiguation/) · 83 works ## Cite Cite as: Sudip Kumar Naskar and Sivaji Bandyopadhyay. 2007. JU-SKNSB: Extended WordNet Based WSD on the English All-Words Task at SemEval-1. Proceedings of the Fourth International Workshop on Semantic Evaluations (SemEval-2007). The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{naskar2007ju, title = {{JU-SKNSB: Extended WordNet Based WSD on the English All-Words Task at SemEval-1}}, author = {Sudip Kumar Naskar and Sivaji Bandyopadhyay}, booktitle = {Proceedings of the Fourth International Workshop on Semantic Evaluations (SemEval-2007)}, year = {2007}, doi = {10.3115/1621474.1621517}, url = {https://doi.org/10.3115/1621474.1621517} } ``` - [← previous Contrasting the Automatic Identification of Two Discourse Markers in Multiparty Dialogues](https://md.research.elamur.ai/free/en/papers/belis-2007-contrasting-automatic-identification-two-discourse/) - [next → SVO triple based Latent Semantic Analysis for recognising textual entailment](https://md.research.elamur.ai/free/en/papers/burek-2007-svo-triple-based-latent-semantic/) - [↑ back to papers of 2007](https://md.research.elamur.ai/free/en/papers/2007/page/2/) ## References 8 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. A. Kilgarriff, and J. Rosenzweig. 2000. Framework and Results for English SENSEVAL. Computers and the Humanities, 34, 15-48. [↗](https://doi.org/10.1023/a:1002693207386) 2. Florentina Vasilescu, Philippe Langlais, and Guy Lapalme. 2004. Evaluating Variants of the Lesk Approach for Disambiguating Words. LREC, Portugal. [find](https://scholar.google.com/scholar?q=Florentina+Vasilescu%2C+Philippe+Langlais%2C+and+Guy+Lapalme%2E+2004%2E+Evaluating+Variants+of+the+Lesk+Approach+for+Disambiguating+Words%2E+LREC%2C+Portugal%2E) 3. G. Ramakrishnan, B. Prithviraj, and P. Bhattacharyya. 2004. A Gloss Centered Algorithm for Word Sense Disambiguation. Proceedings of the ACL SEN-SEVAL 2004, Barcelona, Spain, 217-221. [find](https://scholar.google.com/scholar?q=G%2E+Ramakrishnan%2C+B%2E+Prithviraj%2C+and+P%2E+Bhattacharyya%2E+2004%2E+A+Gloss+Centered+Algorithm+for+Word+Sense+Disambiguation%2E+Proceedings+of+the+ACL+SEN%2DSEVAL+2004%2C+Barcelona%2C+Spain%2C+217%2D221%2E) 4. M. Lesk. 1986. Automatic sense disambiguation using machine readable dictionaries: How to tell a pine cone from a ice cream cone. Proceedings of SIGDOC ’86. [find](https://scholar.google.com/scholar?q=M%2E+Lesk%2E+1986%2E+Automatic+sense+disambiguation+using+machine+readable+dictionaries%3A+How+to+tell+a+pine+cone+from+a+ice+cream+cone%2E+Proceedings+of+SIGDOC+%E2%80%9986%2E) 5. P. Edmonds. 2002. SENSEVAL : The Evaluation of Word Sense Disambiguation Systems, ELRA Newsletter, Vol. 7, No. 3. [find](https://scholar.google.com/scholar?q=P%2E+Edmonds%2E+2002%2E+SENSEVAL+%3A+The+Evaluation+of+Word+Sense+Disambiguation+Systems%2C+ELRA+Newsletter%2C+Vol%2E+7%2C+No%2E+3%2E) 6. S. Banerjee. 2002. Adapting the Lesk Algorithm for Word Sense Disambiguation to WordNet. MS Thesis, University of Minnesota. [find](https://scholar.google.com/scholar?q=S%2E+Banerjee%2E+2002%2E+Adapting+the+Lesk+Algorithm+for+Word+Sense+Disambiguation+to+WordNet%2E+MS+Thesis%2C+University+of+Minnesota%2E) 7. S. Banerjee, and T. Pedersen. 2002. An Adapted Lesk Algorithm for Word Sense Disambiguation Using WordNet. CICLing, Mexico. [find](https://scholar.google.com/scholar?q=S%2E+Banerjee%2C+and+T%2E+Pedersen%2E+2002%2E+An+Adapted+Lesk+Algorithm+for+Word+Sense+Disambiguation+Using+WordNet%2E+CICLing%2C+Mexico%2E) 8. S. Harabagiu, G. Miller, and D. Moldovan. 1999. WordNet2 - a morphologically and semantically enhanced resource. Proceedings of SIGLEX-99, Univ of Mariland. 1-8. [find](https://scholar.google.com/scholar?q=S%2E+Harabagiu%2C+G%2E+Miller%2C+and+D%2E+Moldovan%2E+1999%2E+WordNet2+%2D+a+morphologically+and+semantically+enhanced+resource%2E+Proceedings+of+SIGLEX%2D99%2C+Univ+of+Mariland%2E+1%2D8%2E) *Updated 1 November 2026*