## Logical Models in Information Retrieval: Introduction and Overview (1998)

### Cached

### Download Links

- [www.dcs.qmw.ac.uk]
- [ir.dcs.gla.ac.uk]
- [www.dcs.gla.ac.uk]
- DBLP

### Other Repositories/Bibliography

Venue: | Information Processing & Management |

Citations: | 27 - 5 self |

### BibTeX

@INPROCEEDINGS{Lalmas98logicalmodels,

author = {Mounia Lalmas},

title = {Logical Models in Information Retrieval: Introduction and Overview},

booktitle = {Information Processing & Management},

year = {1998},

pages = {34--1}

}

### OpenURL

### Abstract

The use of logic to model the information retrieval process has become an established research area. Nevertheless, many people in the information retrieval community do not yet appreciate the work performed in this area, mainly because they do not understand logical formalisms, and hence cannot see the connection between logic and information retrieval. This paper aims at resolving the problem. It introduces the formalisms used in logical models for information retrieval, shows the use of logic to build the models, and presents a brief overview of some of the current logical models in information retrieval. 2 1 INTRODUCTION It has been argued that current information retrieval (IR) models offer only simplistic and specific representations of information (Chiaramella and Chevallet, 1992, Nie, 1990, van Rijsbergen, 1989). There is, therefore, a need for the development of a new formalism able to model IR systems in a more generic manner, hence capturing information as it appear...

### Citations

2456 |
A Mathematical Theory of Evidence
- Shafer
- 1976
(Show Context)
Citation Context ...ropriate frameworks are those that are based on what is described as theories of uncertainty, for example, Probability Theory (Pearl, 1988), Bayesian networks (Duda et al., 1976), Theory of Evidence (=-=Shafer, 1976-=-), and Fuzzy Set Theory (Zadeh, 1987). 4.7 Conclusion Various characteristics of a logic that encompasses important attributes of the IR process have been discussed. Classical Logic does not capture w... |

1484 |
A logic for default reasoning
- Reiter
- 1980
(Show Context)
Citation Context ...t. The correspondence cannot be expressed in Classical logic. Examples of logics that capture uncertainty in the inference process are those dealing with non-monotonic reasoning (Krauss et al., 1990, =-=Reiter, 1980-=-). However, the approach often adopted in logical IR modeling is to numerically express the correspondence between relevance and uncertainty. The correspondence is done by attaching a number to each i... |

754 |
An Introduction to Modal Logic
- Hughes, Cresswell
- 1968
(Show Context)
Citation Context ... initial stage of representation of information, knowledge, or belief that may vary by acquiring new information, or by refuting old information. Other logics dealing with partiality are Modal Logic (=-=Hughes and Cresswell, 1968-=-) and Situation Theory (Devlin, 1991). In these, different stages of representation are fixed, and acquiring new information consists of going from one stage to another. * The terminology is not to be... |

669 | Light linear logic
- Girard
- 1998
(Show Context)
Citation Context ... tautology. Therefore, the fact that an information item appears many times, thus showing its significance, cannot be captured. An example of a logic that refutes the above tautology is linear logic (=-=Girard, 1987-=-). Another way to capture significance is to use a weighting mechanism in parallel to the logic. Many conventional IR models use (successfully) weighting mechanisms, and hence empirical results alread... |

643 |
Knowledge in Flux: Modeling the Dynamics of Epistemic States
- Gärdenfors
- 1998
(Show Context)
Citation Context ...and unrelated entities. Therefore, Classical Logic cannot capture partiality. There are many logics that deal with partiality. For example, Three-valued Logic (Kleene, 1967), Belief Revision Systems (=-=Gardenfors, 1988-=-) and Data Semantics (Landman, 1986) all of which postulate an initial stage of representation of information, knowledge, or belief that may vary by acquiring new information, or by refuting old infor... |

560 | Nonmonotonic reasoning, preferential models and cumulative logics
- Kraus, Magidor
- 1990
(Show Context)
Citation Context ... relevant the document. The correspondence cannot be expressed in Classical logic. Examples of logics that capture uncertainty in the inference process are those dealing with non-monotonic reasoning (=-=Krauss et al., 1990-=-, Reiter, 1980). However, the approach often adopted in logical IR modeling is to numerically express the correspondence between relevance and uncertainty. The correspondence is done by attaching a nu... |

469 |
Knowledge and the Flow of Information
- Dretske
- 1981
(Show Context)
Citation Context ... country skiing” may be relevant to a query about “Scandinavian sports”, even if the latter is not explicit in the document. The phenomenon of information containment defines the flow of informa=-=tion (Dretske, 1981-=-). In general, the flow of information is defined between two objects as the information an object contains about another object. If the first object is a text, an image, or a video and the second obj... |

275 |
Introduction to Montague Semantics
- Dowty, Wall, et al.
- 1981
(Show Context)
Citation Context ...ced by 2. Such a substitution is not always correct if 0 represents a polysemic word. Hence, Classical Logic cannot handle intensionality. A logic that deals with intensionality is Intensional Logic (=-=Dowty et al., 1981-=-, van Benthem, 1985). There, the meaning (semantics) of an information item, in one or several contexts, is represented with indices, expressing, for example, times and locations. 12s4.4 Partiality Ma... |

250 |
Logic and Information
- Devlin
- 1991
(Show Context)
Citation Context ...cument contains information pertinent to the query. Logics concerned with the representation of informative relationships are Conditional Logic (Harper et al., 1981, Nute, 1980) and Situation Theory (=-=Devlin, 1991). In -=-the former, an implication � 3 � is only evaluated in those (closest) worlds (a notion explained in section 5.1) in which � is true. In the second, an implication, referred to as a constraint, l... |

217 |
The Situation in Logic
- Barwise
- 1989
(Show Context)
Citation Context ...ow when the implicit information becomes available, for example, by means of the flow of information (explained in section 4.5). This characteristic is referred to as the partiality * of information (=-=Barwise, 1989-=-, Landman, 1986). The representation of partiality in Classical Logic would need to express that the truth value of a formula may not be known at some point, but may become known at some later stage. ... |

183 | A non-classical logic for Information Retrieval - Rijsbergen - 1986 |

172 | Conceptual Structures - Sowa - 1984 |

94 | A model of information retrieval based on a terminological logic
- Meghini, Sebastiani, et al.
- 1993
(Show Context)
Citation Context ... flavored knowledge representation. The primary syntax starts with terms, which are either individuals or relations. Concepts are defined on top of those. The logic was used in the model proposed in (=-=Meghini et al., 1993-=-, Sebastiani, 1994). Documents are represented by individual constants, whereas a class of documents is represented as a concept. The fact that a particular individual is an instance of a concept is w... |

93 |
Probabilities of Conditionals and Conditional Probabilities
- Lewis
- 1976
(Show Context)
Citation Context ...represent medical documents and software components. The problem, however, is the automatic construction of graphs from documents or queries. 5.2 Models based on Imaging Imaging (Harper et al., 1981, =-=Lewis, 1976, Nute, 1980) -=-is also a framework based on the notion of possible-worlds. The truth value of the implication � 3 � in a world � depends on two cases. If � is true in �, then � 3 � is true (false) in t... |

69 |
Stratified Information Disclosure: a synthesis between Hypermedia and Information Retrieval” Phd thesis, Katholieke Universiteit Nijmegen, The Netherlands
- Bruza
- 1993
(Show Context)
Citation Context ...r derivation relations have been proposed which are more flexible than that of Classical Logic. An example is one that captures plausible reasoning, which has been used to develop a logical IR model (=-=Bruza, 1993-=-, Bruza and van der Weide, 1992). The purpose of the model is to capture syntactically related information. The documents and queries are represented by index expressions defined upon noun-phrases. Fo... |

65 |
Semantical analysis of modal logic. Zeitschrift fur Mathematische Logik und Grudlagen der
- Kripke
- 1963
(Show Context)
Citation Context ...fer to the cited literature for the details of the various models. 5.1 Models based on Modal Logic and conceptual graphs Modal Logic (Hughes and Cresswell, 1968) adopts the notion of possible–worlds=-= (Kripke, 1963) -=-which correspond to the interpretations defined in section 2, but which are connected † Fuzzy Logic (Zadeh, 1987) has been used in IR (for example, in (Kracker, 1991)). However, its use was quantita... |

59 | A probabilistic terminological logic for modelling information retrieval
- Sebastiani
- 1994
(Show Context)
Citation Context ...presentation. The primary syntax starts with terms, which are either individuals or relations. Concepts are defined on top of those. The logic was used in the model proposed in (Meghini et al., 1993, =-=Sebastiani, 1994-=-). Documents are represented by individual constants, whereas a class of documents is represented as a concept. The fact that a particular individual is an instance of a concept is written as an asser... |

47 |
An information retrieval model based on modal logic
- Nie
- 1989
(Show Context)
Citation Context ... proposition is with respect to a possible world, and may involve the evaluation of the truth of the proposition or others in connected worlds. Modal Logic was used to develop a logical model for IR (=-=Nie, 1989, Nie, 199-=-0). Documents are worlds, and queries are formulae. A document represented by a world � is relevant to a query represented by a formula � if � is true in �, or if it is true in a world � H a... |

47 | sense disambiguation and information retrieval - Word - 1994 |

45 | Stratified hypermedia structures for information disclosure - Bruza, Weide - 1992 |

45 | Topics in Conditional Logic - Nute - 1980 |

44 | Investigating aboutness axioms using information fields
- Bruza, Huibers
- 1994
(Show Context)
Citation Context ...riately and powerfully defined within a logic (Barwise, 1993, Crossley et al., 1972). The reasons are threefold. Firstly, logical models are believed to be more general than other existing IR models (=-=Bruza and Huibers, 1994-=-, Nie, 1990). Indeed, some logical models are able to represent within a uniform framework various features of IR systems, such as the semantics of information (Nie, 1990), hypermedia objects (Muller ... |

41 |
About retrieval model and logic
- Chiaramella, Chevallet
- 1992
(Show Context)
Citation Context ... current logical models in information retrieval. 2s1 INTRODUCTION It has been argued that current information retrieval (IR) models offer only simplistic and specific representations of information (=-=Chiaramella and Chevallet, 1992-=-, Nie, 1990, van Rijsbergen, 1989). There is, therefore, a need for the development of a new formalism able to model IR systems in a more generic manner, hence capturing information as it appears in a... |

39 |
An Axiomatic Theory for Information Retrieval
- Huibers
- 1996
(Show Context)
Citation Context ...elopment of a new formalism able to model IR systems in a more generic manner, hence capturing information as it appears in an IR system, and also in any of its inherent forms. It has been suggested (=-=Huibers, 1996-=-, Nie, 1990, van Rijsbergen, 1986a, van Rijsbergen, 1986b) that such formalisms can be both appropriately and powerfully defined within a logic (Barwise, 1993, Crossley et al., 1972). The reasons are ... |

38 | Information Retrieval by Logical Imaging - Crestani, Rijsbergen - 1995 |

37 | A study of probability kinematics in information retrieval - Crestani, Rijsbergen - 1998 |

35 | On the role of logic in information retrieval
- Sebastiani
- 1998
(Show Context)
Citation Context ... correspondence between the inference system and the model system does not exist. † An excellent discussion on the difference between truth and validity of � 3 � in the context of IR can be foun=-=d in (Sebastiani, 1995). 8sto asserting � �a �, ��-=-�a � 3 �, � – � or – � 3 �. Here, for simplicity, � �a � is used, which consists of establishing whether any model of � is a model of �. Here is a working example. Suppose th... |

26 | Information retrieval as counterfactual
- Nie, Lepage, et al.
- 1996
(Show Context)
Citation Context ...mantics of information (Nie, 1990), hypermedia objects (Muller and Kutschekmanesch, 1995, Thiel and Muller, 1996) and structured multimedia documents (Chiaramella et al., 1996), the user’s knowledge=-= (Nie et al., 1996-=-) and the nature of IR agents (Huibers and van Linder, 1996). This is not to say that these features were not represented in other IR models, but when they were, often it was not as part of the model,... |

25 | Rijsbergen. The troubles with using a logical model of ir on a large collection of documents. experimenting retrieval by logical imaging on trec. in proceedings of the fourth text retrieval conference (trec-4 - Crestani, Ruthven, et al. - 1995 |

23 |
Fuzzy Sets and Applications: Selected Papers
- Zadeh
- 1987
(Show Context)
Citation Context ...e based on what is described as theories of uncertainty, for example, Probability Theory (Pearl, 1988), Bayesian networks (Duda et al., 1976), Theory of Evidence (Shafer, 1976), and Fuzzy Set Theory (=-=Zadeh, 1987-=-). 4.7 Conclusion Various characteristics of a logic that encompasses important attributes of the IR process have been discussed. Classical Logic does not capture well, and in some instances, does not... |

14 | Towards a theory of information, the status of partial objects in semantics. Dordrecht: Foris - Landman - 1986 |

11 |
A Fuzzy Concept Network Model and its Applications
- Kracker
- 1992
(Show Context)
Citation Context ...ts the notion of possible–worlds (Kripke, 1963) which correspond to the interpretations defined in section 2, but which are connected † Fuzzy Logic (Zadeh, 1987) has been used in IR (for example, =-=in (Kracker, 1991)). -=-However, its use was quantitative (it was used as an alternative to Probability Theory, for example). The use of “quantitative logic” in IR is not described in this paper. 17sto each other via an ... |

11 | A new theoretical framework for information retrieval[C - van - 1986 |

10 | Situation Theory and Dempster-Shafer's Theory of Evidence for Information Retrieval - Lalmas, Rijsbergen - 1993 |

8 |
Un modele logique de recherche d'information applique au formalisme des graphes conceptuels. Le prototype ELEN et son experimentation sur un corpus de composants logiciels
- Chevallet
- 1992
(Show Context)
Citation Context ... models in information retrieval. 2s1 INTRODUCTION It has been argued that current information retrieval (IR) models offer only simplistic and specific representations of information (Chiaramella and =-=Chevallet, 1992-=-, Nie, 1990, van Rijsbergen, 1989). There is, therefore, a need for the development of a new formalism able to model IR systems in a more generic manner, hence capturing information as it appears in a... |

7 | Formalising Intelligent Information Retrieval Agents - Huibers, Linder - 1996 |

7 |
Un Mod`ele de Logique G'en'erale pour les Systemes de Recherche d'Informations. Application au Prototype RIME
- Nie
- 1990
(Show Context)
Citation Context ...ation retrieval. 2s1 INTRODUCTION It has been argued that current information retrieval (IR) models offer only simplistic and specific representations of information (Chiaramella and Chevallet, 1992, =-=Nie, 1990-=-, van Rijsbergen, 1989). There is, therefore, a need for the development of a new formalism able to model IR systems in a more generic manner, hence capturing information as it appears in an IR system... |

6 |
An AI view of the treatment of uncertainty
- Saffioti
- 1987
(Show Context)
Citation Context ... the evaluation qualitative as opposed to quantitative and to use concurrently a theory of uncertainty to embody partial relevance and uncertainty. The second of these directions is not uncommon, as (=-=Saffioti, 1987) -=-says: “Many of these solutions [of representing uncertainty] share the attitude of viewing the knowledge and the uncertainty about it as two different entities, and so treating them by means of two ... |

6 |
Why was this Item Retrieved? New Ways to Explore Retrieval Results
- Thiel, Muller
- 1996
(Show Context)
Citation Context ...logical models are able to represent within a uniform framework various features of IR systems, such as the semantics of information (Nie, 1990), hypermedia objects (Muller and Kutschekmanesch, 1995, =-=Thiel and Muller, 1996) -=-and structured multimedia documents (Chiaramella et al., 1996), the user’s knowledge (Nie et al., 1996) and the nature of IR agents (Huibers and van Linder, 1996). This is not to say that these feat... |

5 |
An Overview of Information Retrieval
- Smeaton
- 1996
(Show Context)
Citation Context ...ss important than others, and hence their implementations can be simplified. Finally, it may also identify which logic leads to the most effective IR systems. In conclusion, the following quote from (=-=Smeaton, 1996) -=-summarizes my beliefs in such work: “While currently much of this work [the use of logic in IR modeling] can be 26sregarded as an interesting theoretical exercise because of the small scale of actua... |

1 |
Subjective Bayesian methods for rules-based inference systems
- Duda, Hart, et al.
- 1976
(Show Context)
Citation Context ...s are used in parallel to the logic. Appropriate frameworks are those that are based on what is described as theories of uncertainty, for example, Probability Theory (Pearl, 1988), Bayesian networks (=-=Duda et al., 1976-=-), Theory of Evidence (Shafer, 1976), and Fuzzy Set Theory (Zadeh, 1987). 4.7 Conclusion Various characteristics of a logic that encompasses important attributes of the IR process have been discussed.... |

1 | A Manual of Intentional Logic - Benthem - 1985 |