Kathleen McKeown, Judith Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin. 1999. Towards multidocument summarization by reformation: Progress and prospects. In Proceedings of American Association for Artificial Intelligence 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....use those terms present in the query. This meant that for the document portion of the ranking value, we only summed over query terms. The results, displayed in Figure 4, show performance under this scheme is better in all areas of the curve than the Poisson model, as well as comparable to tf.idf [18], performing slightly worse in areas of high precision, and better in areas of low precision. 0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 Precision Tf.idf New Model Poisson Figure 4: The hyper learned model on short queries, compared with the Poisson model and tf.idf. 8. CONCLUSIONS AND ....

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulation: Progress and prospects. In AAAI, 1999.

....in previous work[10] These comparisons show that ours perform as well as or better at summarization and finding relevant documents than these alternatives. From the perspective of multi document summarization, most automatic techniques extract sentences or portions of sentences from documents[13, 15, 19], and then string them together to form a summary. These techniques are only suitable for a small number of documents and are incapable of handling the larger document sets that hierarchies can. 7. CONCLUSIONS AND FUTURE WORK This paper describes our design and implementation of hierarchies for ....

K. McKeown, J. Klavans, V. Hatzivzssiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulation: Progress and prospects. In Proceedings of the Sixteenth National Conference on Artificial Intelligence, pages 453--460, 1999.

....0004 3702 02 see front matter 2002 Elsevier Science B.V. All rights reserved. PII: S0004 3702(02)00222 9 catenation of textual segments does not yield coherent outputs. Recently, a number of researchers have started to address the problem of generating coherent summaries: McKeown et al. [26], Barzilay et al. 3] Jing and McKeown [15] Barzilay et al. 2] and Marcu and Gerber [25] in the context of multidocument summarization; and Mani et al. 22] in the context of revising single document extracts. The approaches proposed by Witbrock and Mittal [29] Banko et al. 1] Berger and ....

.... A sentence compression module can also be used to provide audio scanning services for the blind [13] and faster access to the web from PDA devices [7] In general, since all systems aimed at producing coherent abstracts often implement manually written sets of sentence compression rules [3,22, 26], it is likely that a good sentence compression module would impact the overall quality of these systems as well. This becomes particularly important for text genres that use long sentences. Previous rule based work addressing sentence compression includes Jing and Mckeown [15] Mahesh [21] ....

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, E. Eskin, Towards multidocument summarization by reformulation: Progress and prospects, in: Proceedings of AAAI-99, Orlando, FL, 1999, pp. 453--460.

....applied statistical techniques (frequency analysis, variance analysis, etc. to linguistic units suchastokens, names, anaphora, etc. e.g. 27, 19, 9, 18, 2] Other approaches include the utility of discourse structure [14] the combination of information extraction and language generation [11, 17, 24, 21, 16], and using machine learning to find patterns in text [28, 4, 26] Several researchers have extended various aspects of the single document approaches to look at multi documentsummarization [13, 21, 3, 7, 15] These include comparing templates filled in by extracting information using ....

....167 respect to an already provided summary, 4) the developmentofanevent or subtopic of an event (e.g. death tolls) over time, and (5) a comparativedevelopmentofanevent. Naturally, an ideal multi document summary would include natural language generation to create cohesive readable summaries [21, 16]. Our focus is on fast, domain independent summaries, which is currently beyond the scope of natural language processing techniques. 5. SYSTEM DESIGN In the previous sections we discussed the requirements and types of multi document summarization systems. This section discusses our current ....

Kathleen McKeown, Judith Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin. Towards Multidocument Summarization by Reformulation: Progress and Prospects. In Proceedings of AAAI-99, pages 453--460, Orlando, FL, July 1999.

....(1997) also relate pairs of documents to each other showing similarities and differences. In addition, work by McKeown and Radev (McKeown and Radev 1995; Radev and McKeown 1998) relies on an assumed system filling and selecting predefined templates used for the final summary. Later work by McKeown et al. 1999) breaks documents into paragraph based units. These units are compared to each other to identify similar and dissimilar passages. A graph based one pass clustering algorithm is applied, using the similarity metric, to identify common topics themes. Instead of picking a representative sentence from ....

....Summarization: Methodologies and Evaluations summary and detect redundancy. Similarly to our approach, clusters are formed and a representative segment is presented to the user. We are currently not aware of any formal evaluation of multi document summarization other than work by described in McKeown et al. 1999) and Stein et al. 2000) However, both evaluations are system specific and evaluate certain aspects. The first system focuses on three system components: the similarity metric (evaluated using TDT data) the theme phrase detection approach and the sentence generation capability. The second system ....

MCKEOWN, KATHLEEN R., JUDITH L. KLAVANS, REGINA BARZILAY and ELEAZAR ESKIN. 1999. Towards Multidocument Summarization by Reformulation: Progress and Prospects. AAAI '99, 453 -- 460.

....(1997) also relate pairs of documents to each other showing similarities and differences. In addition, work by McKeown and Radev (McKeown and Radev 1995; Radev and McKeown 1998) relies on an assumed system filling and selecting predefined templates used for the final summary. Later work by McKeown et al. 1999) breaks documents into paragraph based units. These units are compared to each other to identify similar and dissimilar passages. A graph based one pass clustering algorithm is applied, using the similarity metric, to identify common topics themes. Instead of picking a representative sentence from ....

....the case of our system we do not have pre specified perfect output, nor can we easily measure improved task performance since this is not directly linked to the clustering technique. We are currently not aware of any formal evaluation of multidocument summarization other than work described in McKeown et al. 1999). This evaluation is system specific and focuses in particular on three system components: the similarity metric (evaluated using TDT data) the theme phrase detection approach and the sentence generation capability. 6. Discussion, Future Work Our future work will be extending our current system ....

McKeown, Kathleen R., Judith L. Klavans, Regina Barzilay and Eleazar Eskin. 1999. Towards Multidocument Summarization by Reformulation: Progress and Prospects. AAAI '99, 453 -- 460.

....the most important textual segments is only half of what a summarization system needs to do because, in most cases, the simple catenation of textual segments does not yield coherent outputs. Recently, a number of researchers have started to address the problem of generating coherent summaries: McKeown et al. 1999), Barzilay et al. 1999) and Jing and McKeown (1999) in the context of multidocument summarization; Mani et al. 1999) in the context of revising single document extracts; and Witbrock and Mittal (1999) in the context of headline generation. The approach proposed by Witbrock and Mittal (1999) is ....

....a summarization system needs to do because, in most cases, the simple catenation of textual segments does not yield coherent outputs. Recently, a number of researchers have started to address the problem of generating coherent summaries: McKeown et al. 1999) Barzilay et al. 1999) and Jing and McKeown (1999) in the context of multidocument summarization; Mani et al. 1999) in the context of revising single document extracts; and Witbrock and Mittal (1999) in the context of headline generation. The approach proposed by Witbrock and Mittal (1999) is the only one that applies a probabilistic model ....

[Article contains additional citation context not shown here]

McKeown, K.; Klavans, J.; Hatzivassiloglou, V.; Barzilay, R.; and Eskin, E. 1999. Towards multidocument summarization by reformulation: Progress and prospects. In Proceedings of the Sixteenth National Conference on Artificial Intelligence (AAAI--99).

No context found.

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulatin: Progress and prospects. In Proceedings of the Seventeenth National Conference on Arti cial Intelligence, 1999.

No context found.

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. 1999. Towards multidocument summarization by reformulation: Progress and prospects. In Proc. of the 17th National Conference on Artificial Intelligence.

No context found.

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulation: Progress and prospects. In Proc. of the AAAI Conf., 1999.

No context found.

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulatin: Progress and prospects. In Proceedings of the Seventeenth National ConferenceonArti cial Intelligence, 1999.

....The training and test sets for multidocument summarization developed by NIST for the Document Understanding Conference demonstrate the large number of possible variations. Summarization of document clusters that are tightly focused on a single event is the research focus of Columbia s MultiGen [11], HLT 02, San Diego, Calif. USA . a summarization system that uses information fusion and similarities across input articles. However, since many of the documents in the DUC training corpus in 2001 were only loosely connected, we needed to develop an alternative summarization strategy. Our ....

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformation: Progress and prospects. In Proceedings of American Association for Artificial Intelligence 1999.

....Newsblaster is a system developed at Columbia to provide news updates on a daily basis; it crawls news sites, filters out news from non news (e.g. ads) groups news into stories on the same event, and generates a summary of each event. Summaries are generated using the Columbia Summarizer [8, 7, 13], which was evaluated in the Document Understanding Conference (DUC) in 2001. News is grouped into stories on the same event using a Topic Detection and Tracking (TDT) style system developed at Columbia [4] Unlike other TDT systems, Columbia s uses a learned, weighted combination of features to ....

....ships. The time span covered is unpredictable, but longer than in the single event case. Other clusters contain even more loosely related documents and do not fit any of the categories above. To summarize documents on the same event, the Columbia summarizer uses an enhanced version of MultiGen [2, 8]. MultiGen integrates machine learning and statistical techniques to identify similar sentences (set set of similar sentences is called a themes) across the input articles [5, 6] It then uses an alignment of parse trees to find the intersection of similar phrases within sentences [2] It orders ....

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulatin: Progress and prospects. In Proceedings of the Seventeenth National Conference on Artificial Intelligence, 1999.

....summarization strategy that can be adapted to documents of different types, including biographies and multiple weakly related events. To summarize documents on the same event, the Columbia summarizer uses an enhanced ver sion of MultiGen [Barzilay et al. 1999; Hatzivas siloglou et al. 1999; McKeown et al. 1999; Barzi lay et al. 2001; Hatzivassiloglou et al. 2001] for biographical documents, it uses an alternate sys tem, DEMS (Dissimilarity Engine for Multidocument Summarization) tuned to the biographical task; and for sets of loosely similar documents, it uses DEMS with a more general ....

Kathleen R. McKeown, Judith L. Klawns, Vasileios Hatziwssiloglou, Regina Barzilay, and Eleazar Eskin. Towards Multidocument Summarization by Reformulation: Progress and Prospects. In Proceedings of the Seventeenth National Conference on Artificial Intelligence (AAAI-99), pages 453-460, Orlando, Florida, July 1999. American Association for Artificial Intelligence.

....5, six topics were aligned. Given a seven sentence synopsis, Symptoms would receive two sentences whereas Variants would receive only one. Once a topic receives a sentence allotment, we must choose the sentences to represent it. To perform this task, we utilize a sentence clustering technique [7, 3] that takes as input a set of sentences and organizes them into clusters based on their sentential similarity. For each topic, we run the clustering program on the sentences of the topic instances, producing clusters of similar sentences as output. We now chose a single sentence to represent each ....

K. R. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulation: Progress and prospects. In "Proceedings of the Seventeenth National Conference on Artificial Intelligence (AAAI-99)". ACL, July 1999.

No context found.

Kathleen R. McKeown, Judith L. Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin. 1999. Towards multidocument summarization by reformulation: Progress and prospects. In Proceedings of the Seventeenth National Conference on Artificial Intelligence (AAAI-99), pages 453--460, Orlando, Florida, July.

....output for each document, detecting similarities and differences, and tracking how that topic, as represented by the SNPs, changes over time. LinkIT has also been used in a paragraph level similarity detection component of a multiple document summarization system (Hatzivassiloglou et al. 1999; McKeown et al. 1999). The output from LinkIT could also be used as the input for a term variant finder, such as FASTR (Jacquemin 1999) It would be possible to use LinkIT on a selection of documents that has been shown likely to be relevant by some other method, in order to make more fine distinctions between the ....

McKeown, Kathleen R., Judith L. Klavans, Vasileios Hatzivassiloglou, Regina Barzilay and Eleazar Eskin, (1999). Towards Multidocument Summarization by Reformulation: Progress and Prospects. In Proceedings of the Sixteenth National Conference on Artificial Intelligence AAAI1999. Orlando, Florida.

No context found.

Kathleen R McKeown, Judith Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin. 1999. Towards multidocument summarization by reformulation: Progress and prospects. submitted.

No context found.

Kathleen R McKeown, Judith Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin. 1999. Towards multidocument summarization by reformulation: Progress and prospects. submitted.

.... query results, hence providing users with an overview of the results that is easier to understand and process than a flat list of documents (see, e.g. 7] It can also form the basis for further processing of the documents once they have been organized in topical groups, such as summarization [11]. Clustering is also a key component of DARPA s ongoing Topic Detection and Tracking(TDT) initiative, which completed its second phase (TDT2) in early 1999. 1 The goal of the TDT initiative is to provide benchmarks for comparing systems that 1 See http: www.itl.nist.gov iaui 894.01 tdt98 ....

K. McKeown, J. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. Eskin. Towards multidocument summarization by reformulation: Progress and prospects. In Proceedings of the 17th National Conference on Artificial Intelligence (AAAI-99), pages 453--460, Orlando, Florida, July 1999.

No context found.

Kathleen McKeown, Judith Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin. 1999. Towards multidocument summarization by reformation: Progress and prospects. In Proceedings of American Association for Artificial Intelligence 1999.

No context found.

K. McKeown, J. L. Klavans, V. Hatzivassiloglou, R. Barzilay, and E. E. Towards multidocument summarization by reformulation: Progress and prospects. In Proceedings of AAAI-99.

No context found.

McKeown, K. and J. Klavans, V. Hatzivassiloglou, R. Barzilay, E. Eskin, 1999. "Towards multidocument summarization by reformulation: Progress and prospects." In Proceedings of AAAI-99, Orlando, Fl., pp. 453-60.

No context found.

McKeown, K., Klavans, J., Hatzivzssiloglou, V., Barzilay, R., and Eskin, E. Towards multidocument summarization by reformulation: Progress and prospects. In Proceedings of the Sixteenth National Conference on Artificial Intelligence (1999), pp. 453--460.

No context found.

Kathleen McKeown, Judith Klavans, Vasileios Hatzivassiloglou, Regina Barzilay, and Eleazar Eskin, Towards Multidocument Summarization by Reformulation: Progress and Prospects, In Proceedings of AAAI'99, Orlando, FL, July 1999.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC