Valentina Franzoni

PMING Distance: A Collaborative Semantic Proximity Measure

One of the main problems that emerges in the classic approach to semantics is the difficulty in acquisition and maintenance of ontologies and semantic annotations. On the other hand, the flow of data and documents which are accessible from the Web is continuously fueled by the contribution of millions of users who interact digitally in a collaborative way. Search engines, continually exploring the Web, are therefore the natural source of information on which to base a modern approach to semantic annotation. A promising idea is that it is possible to generalize the semantic similarity, under the assumption that semantically similar terms behave similarly, and define collaborative proximity measures based on the indexing information returned by search engines. In this work PMING, a new collaborative proximity measure based on search engines, which uses the information provided by search engines, is introduced as a basis to extract semantic content. PMING is defined on the basis of the best features of other state-of-the-art proximity distances which have been considered. It defines the degree of relatedness between terms, by using only the number of documents returned as result for a query, then the measure dynamically reflects the collaborative change made on the web resources. Experiments held on popular collaborative and generalist engines (e.g. Flickr, Youtube, Google, Bing, Yahoo Search) show that PMING outperforms state-of-the-art proximity measures (e.g. Normalized Google Distance, Flickr Distance etc.), in modeling contexts, modeling human perception, and clustering of semantic associations.

Collective evolutionary concept distance based query expansion for effective web document retrieval

In this work several semantic approaches to concept-based query expansion and re-ranking schemes are studied and compared with different ontology-based expansion methods in web document search and retrieval. In particular, we focus on concept-based query expansion schemes where, in order to effectively increase the precision of web document retrieval and to decrease the users’ browsing time, the main goal is to quickly provide users with the most suitable query expansion. Two key tasks for query expansion in web document retrieval are to find the expansion candidates, as the closest concepts in web document domain, and to rank the expanded queries properly. The approach we propose aims at improving the expansion phase for better web document retrieval and precision. The basic idea is to measure the distance between candidate concepts using the PMING distance, a collaborative semantic proximity measure, i.e. a measure which can be computed using statistical results from a web search engine. Experiments show that the proposed technique can provide users with more satisfying expansion results and improve the quality of web document retrieval.

Heuristic semantic walk for concept chaining in collaborative networks

Purpose – In this work, a new general framework is proposed to guide navigation over a collaborative concept network, in order to discover paths between concepts. Finding semantic chains between concepts over a semantic network is an issue of great interest for many applications, such as explanation generation and query expansion. Collaborative concept networks over the web tend to have features such as large dimensions, high connectivity degree, dynamically evolution over the time, which represent special challenges for efficient graph search methods, since they result in huge memory requirements, high branching factors, unknown dimensions and high cost for accessing nodes. The paper aims to discuss these issues. Design/methodology/approach – The proposed framework is based on the novel notion of heuristic semantic walk (HSW). In the HSW framework, a semantic proximity measure among concepts, reflecting the collective knowledge embedded in search engines or other statistical sources, is used as a heurist...

Set Similarity Measures for Images Based on Collective Knowledge

This work introduces a new class of group similarity where different measures are parameterized with respect to a basic similarity defined on the elements of the sets. Group similarity measures are of great interest for many application domains, since they can be used to evaluate similarity of objects in term of the similarity of the associated sets, for example in multimedia collaborative repositories where images, videos and other multimedia are annotated with meaningful tags whose semantics reflects the collective knowledge of a community of users. The group similarity classes are formally defined and their properties are described and discussed. Experimental results, obtained in the domain of images semantic similarity by using search engine based tag similarity, show the adequacy of the proposed approach in order to reflect the collective notion of semantic similarity.

Heuristics for Semantic Path Search in Wikipedia

In this paper an approach based on Heuristic Semantic Walk (HSW) is presented, where semantic proximity measures among concepts are used as heuristics in order to guide the concept chain search in the collaborative network of Wikipedia, encoding problem-specific knowledge in a problem-independent way. Collaborative information and multimedia repositories over the Web represent a domain of increasing relevance, since users cooperatively add to the objects tags, label, comments and hyperlinks, which reflect their semantic relationships, with or without an underlying structure. As in the case of the so called Big Data, methods for path finding in collaborative web repositories require solving major issues such as large dimensions, high connectivity degree and dynamical evolution of online networks, which make the classical approach ineffective. Experiments held on a range of different semantic measures show that HSW lead to better results than state of the art search methods, and points out the relevant features of suitable proximity measures for the Wikipedia concept network. The extracted semantic paths have many relevant applications such as query expansion, synthesis of explanatory arguments, and simulation of user navigation.

Heuristic Semantic Walk

Path search between concepts over a semantic network is an issue of great interest for many applications, such as explanation generation and query expansion. In this study a new approach is proposed, to guide navigation over a collaborative concept network, in order to discover path between concepts. The method uses a semantic heuristic based on proximity measures, which reflects the collective knowledge embedded in search engines. The experiments held on the Wikipedia network and Bing search engine on a range of different semantic measures show that the proposed approach outperforms state of the art search methods.

Improving Link Ranking Quality by Quasi-Common Neighbourhood

Most of the best performing link prediction ranking measures evaluate the common neighbourhood of a pair of nodes in a network, in order to assess the likelihood of a new link. On the other hand, the same zero rank value is given to node pairs with no common neighbourhood, which usually are a large number of potentially new links, thus resulting in very low quality overall link ranking in terms of average edit distance to the optimal rank. In this paper we introduce a general technique for improving the quality of the ranking of common neighbours-based measures. The proposed method iteratively applies any given ranking measure to the quasi-common neighbours of the node pair. Experiments held on widely accepted datasets show that QCNAA, a quasi-common neighbourhood measure derived from the well know Adamic-Adar (AA), generates rankings which generally improve the ranking quality, while maintaining the prediction capability of the original AA measure.

Semantic context extraction from collaborative networks

A novel method for the automatic online extraction of contexts from collaborative explanation network is introduced. The method explore an unknown online collaborative network in order to find multiple explanatory paths between seed concepts. The exploration is driven by an online randomized walk informed by a heuristics based on semantic proximity measures. A pheromone-like model is then applied to the analysis of the relevance of concepts in multiple explanatory paths in order to extract the relevant contexts. Experiments held on the collaborative network Wikipedia and accepted datasets show that the proposed method is able to determine contexts with high degree of relevance which outperforms other methods. The methodology have general aim and it can be easily extended to other online collaborative networks and to non-textual domains.

Web-based similarity for emotion recognition in web objects

In this project we propose a new approach for emotion recognition using web-based similarity (e.g. confidence, PMI and PMING). We aim to extract basic emotions from short sentences with emotional content (e.g. news titles, tweets, captions), performing a web-based quantitative evaluation of semantic proximity between each word of the analyzed sentence and each emotion of a psychological model (e.g. Plutchik, Ekman, Lovheim). The phases of the extraction include: text preprocessing (tokenization, stop words, filtering), search engine automated query, HTML parsing of results (i.e. scraping), estimation of semantic proximity, ranking of emotions according to proximity measures. The main idea is that, since it is possible to generalize semantic similarity under the assumption that similar concepts co-occur in documents indexed in search engines, therefore also emotions can be generalized in the same way, through tags or terms that express them in a particular language, ranking emotions. Training results are compared to human evaluation, then additional comparative tests on results are performed, both for the global ranking correlation (e.g. Kendall, Spearman, Pearson) both for the evaluation of the emotion linked to each single word. Different from sentiment analysis, our approach works at a deeper level of abstraction, aiming to recognize specific emotions and not only the positive/negative sentiment, in order to predict emotions as semantic data.

Multi-path traces in semantic graphs for latent knowledge elicitation

In this work an online collaborative semantic network is explored. The method used is based on multi-path traces for extracting latent contextual knowledge, which explores an unknown Semantic proximity measures based on search engines are uses as heuristics to navigate the collaborative network, in order to find multiple random paths representing traces between seed concepts. The exploration is driven by an online randomized walk informed by those heuristics, where the multiple traces model reinforces the most relevant explanatory paths using a pheromone-like approach to elicit latent contexts. Experiments have been held on Wikipedia and on the Word Similarity 353 dataset to evaluate the effectiveness of the method. The general methodology can be easily extended to other online collaborative graphs and to non-textual domains.