WO2014009015A2

WO2014009015A2 - Module having an interactive representation of a search query for a computer on a display unit for a user

Info

Publication number: WO2014009015A2
Application number: PCT/EP2013/002051
Authority: WO
Inventors: Marcus NITSCHE; Andreas NÜRNBERGER; Kathrin SCHELL
Original assignee: Otto-Von Guericke-Universität Magdeburg Technologie-Transfer-Zentrum
Priority date: 2012-07-10
Filing date: 2013-07-10
Publication date: 2014-01-16
Also published as: WO2014009015A3; DE102012013578A1

Abstract

The invention relates to a module having an interactive representation of a search query on a display unit for a user of a computer, comprising: a search module providing an input of a search, based on which the search module carries out a basic search; the display unit optically displaying a search result of the basic search; a selection option, which displays a partial match of the search result and allows a selection of the partial match; a complementary partial match search, which, once the partial match has been selected, automatically executes an additional search by way of the search module, wherein a content of the partial match feeds in an automated manner into the additional search to complement same; and an array of a result of the complementary partial match search on the display unit, which is different from a display of the search result of the search on the display unit. The invention further relates to a method, and to a computer program product.

Description

Module with an interactive representation of a search query for a computer on a display for an operator

Web and patent searches are applications of information retrieval: "Information retrieval, abbreviated to IR, is the finding of material, mostly documents of an unstructured nature, usually text, containing information in large collections, usually on Computers stored, fulfilled.

IR systems are similar to traditional database systems, where IR systems deal with non-structured (well) structured texts in natural language or other semi-structured or unstructured information, as well as retrieving information as opposed to retrieving data, such as the search and indexing process in an IR system as described in (Baeza-Yates & Ribeiro-Neito, 2011).

The standard IR task consists of two independent processes: an index structure has to be set up first in order to enable an efficient search afterwards.

An index is a special data structure to enable fast searching of documents in traditional and digital libraries. The facility begins with a preprocessing step of the document collection, and is subsequently followed by an index construction, for example, by a simple inverted index (Baeza-Yates & Ribeiro-Neto, 2011).

The search starts with a user's search query. This corresponds to the formulation of his need for information. This query is entered, sent and then processed in the IR system. Various steps are involved. This can be accompanied by techniques that include user feedback such as search suggestions, auto-completion or spell-checking. The index structure is then used to retrieve the results that the system retrieves based on the processed query. Fetching as many relevant documents as possible while minimizing the amount of non-relevant documents retrieved is the main task of an IR system. This is not trivial, as relevance is a personal assessment based on context, time, place, or facility, just to name a few influencing factors.

Another critical step is the ranking of the documents found, as it has a major impact on the perceived quality of the retrieval system. In this phase

CONFIRMATION COPY a relevance order is identified. The results can be sorted by date or popularity. Refined metrics can also leverage the context of the user, document collection properties, or similar properties to aid efficient search.

The final step is the presentation of the sorted documents in a way that facilitates the satisfaction of the information needs of the user, especially in the form of a ranking list. Document representatives with different characteristics have been developed to facilitate this task.

This is the general process, for example, based on Boolean retrieval, assuming that a document is either relevant or not. For this case, several evaluation measures have been developed. The most common are Precision and Recall: While Precision refers to the proportion of relevant results among all returned results, Recall indicates the proportion of returned relevant results among all relevant documents in the document collection. The underlying assessments of relevance to each document have mostly been suggested by human specialists.

A search requires many cognitive abilities of the user such as attention, judgment, learning, problem solving, as well as a number of emotional factors. Thus, the information retrieval community refers to insights from psychology to model the user behavior and to design with these user interfaces that can support real user needs.

The starting point when using an IR system is that a user needs to meet an information need. This need is subject to wide variation, and (Shneiderman, Byrd, and Croft, 1997) define it as:

"The perceived need for information causes someone to use an information retrieval system first."

The user is aware of this knowledge gap in a particular context. As retrieval systems primarily address issues that adequately represent this need, the user attempts to expand a query. Nonetheless, a person may or may not be able to describe the right words for their information needs. The query can not only be specified in text form, but also - especially popular with mobile devices - with voice commands. The object of the present invention is to make a computerized search more intuitive.

This object is achieved with a module having the features of claim 1, with a performance of a computer search with the features of claim 9 and with a computer program product with the features of claim 11. Further advantageous embodiments and further developments will become apparent from the following subclaims. However, one or more features of one or more embodiments of the claims as well as the description and the respective figures may be linked to other embodiments. The proposed respective independent claims serve as an attempt at a first presentation of the subject invention, but without their respective formulation is binding. Rather, one or more features of the independent claims may be supplemented by one or more features of the disclosure, but may be substituted.

A module with an interactive representation of a search query for a computer on a display for an operator is proposed. The module preferably has the following:

a search engine which provides an entry for a search and on the basis of which the search module carries out a basic search,

the display which visually displays a search result for the basic search,

- a choice that displays a partial hit of the search result and allows you to select the partial hit,

- A supplementary part hit search, which independently selects the part hit using the research module performs an additional search, with a content of the part hit automatically included in the additional search supplemental, and

an arrangement of a result of the supplementary part-hit search on the display, which differs from an arrangement of the search result of the search on the display.

An interactive presentation should be understood as meaning that the operator is enabled to influence the search and thus the search terms on the one hand. On the other hand, the interactive display also allows that after displaying the search result on the display, the operator can access the search result as a whole or even in parts and edit it further. For example, there is the possibility of following a result, zooming in, selecting a divider result of a resort, discarding a result or otherwise actuating the module. The operation can be done in a conventional manner by means of the mouse but also in other ways, for example by touching a touch screen, by voice activation, by Pointernutzung, by Augenirisdetektierung or by using any other, an activation permitting device. This may include, in particular, screen pens or otherwise.

The display can be a screen, for example a computer screen. So it can be a workplace screen. Also, the display may be part of a smartphone, another telecommunications device, a laptop, a tablet PC or any other device that is permanently installed, for example. Thus, the display can be installed in a vehicle, for example in a car, an aircraft or in an information device, such as those offered in museums and exhibitions for the use of visitors and interested parties with searchable data or files. So it may be part of a navigation device, a radio, a multimedia device or any other device. The display can also be arranged, for example, in a domestic appliance, for example in a refrigerator or else in a house control device or in a surveillance system. Likewise, the display can be part of an industrial plant device, for example a machine tool, a control panel, a control device or the like. The display can also be a television, part of a game console, a camera, a camera or other portable device. The display can also be formed by a projection on a surface, such as in a vehicle, the windscreen, by means of a beamer or other projection devices.

The display can be created in many ways, for example as an LCD display, as an LED display, as an AMOLED display or by other known techniques. A resolution is preferably adjustable, preferably also during operation of the module. As a result, depending on the result found, a suitable display resolution can be set. This is preferably done automatically. Instead of a single display, two or more displays can be coupled together, on which then the respective result is displayed. For example, two monitors can stand side by side, with the display of results going from one monitor to the other.

In the following, the search query is to be understood as a representation of a search. This search can be done by one or more words that can be entered. Example- way this is an input, preferably an input field displayed on the display. The search can also be carried out by other search terms, by search strings, by images, by combinations of different documents and thus in a variety of ways.

The research module is able to accept search jobs, execute them and provide search results. The search module is arranged, for example, on a CPU of a computer or a memory of a computer. The search module can also be arranged on one or more servers. These can be coupled to each other in particular by remote connection. The research module is able to access one or more databases for research. For this purpose, the search module can have one or more search algorithms.

The search result is the retrieved search result shown on the display. The display form can be done in different ways. For example, a large number of individual frames with content can be displayed. Through the frames, the individual results of a result can be distinguished from each other. Even through other, clear structures, individual hits can be presented distinctively from one another as a search result on the display.

Incidentally, a presentation on the display can also include the option of displaying an increased resolution, in particular for the visually impaired. The results of the result can also be made known acoustically, for example. For this purpose, for example, a reading program can be additionally provided. By selecting, the result can then also be provided acoustically by selected respective hit areas from the respective hits. This is for example for the visually impaired but also in vehicles an advantage.

The supplementary partial hit search is carried out automatically after selecting one or more hits from a displayed result. This is automated, for example, again supplemented by means of an algorithm with reference to the selected hit the original search with one or more additional search terms. For this purpose, the content of the selected hit is automatically analyzed. By specifying the analysis type, the search term to be supplemented can be found and supplemented. The partial hit search automatically performs a search in this way and displays the results found as a further result. The arrangement of the result of the partial search result differs from that of the result of the basic search. In this case, the basic search is to be understood as the search from which the partial research was started. According to a preferred embodiment, the arrangement of the two results takes place simultaneously on the display. In particular, due to the different arrangement of the two results, the operator can also see an origin and a search path continuing from the origin. For example, there may be an orthogonal arrangement of result to partial search result. Another two- or even three-dimensional arrangement of result and partial search result is possible. Also a different size, color marking. Sharpness or blur, highlighting or moving in the background arrangement is available as a possibility.

The module in turn can thus be present in various individual components. These individual components may be integrated with each other in a single device, for example a fixed installation which also contains all the necessary data to be searched. However, the individual components can also be present locally separated from one another, but coupled to one another by appropriate connections. In this case, connections can be signal-transmitting connections, for example wireless as well as wired connections. Here, the term of the wired connections is not limited to a metallic wire connection to understand. Rather, fiber optic links or other data transmitting conductors may also be understood. Furthermore, physical connections may also be provided, for example by a housing, by arrangement on a circuit board and the like.

A further embodiment of the proposed module provides that the arrangement of several results of the supplementary partial search on the display is orthogonal to the search result of the base search. Preferably, both are at least partially displayed simultaneously.

Another embodiment provides that the further results of a search are not simply positioned orthogonally to the previous search path, but are always appended only to the right or to the bottom. From the design point of view, this has the advantage of a consistent presentation and a fixed start relative to the respective origin. For example, the user does not have to scroll to the left or the top forever, but roughly follows his search path down to the right. A further embodiment of the module provides that the arrangement of different results of successive searches is displayed in a tree-like structure on the display. This facilitates the overview, but without ignoring the respective starting point. In addition, this, as well as the above representations, allows for zooming in as well as moving away to show more detailed results on different nodes, as well as larger overviews on the other hand.

For example, there is the possibility that the arrangement of at least one result on the display by the operator is displaced. Incidentally, the number or arrangement of search results is preferably not limited by the display. According to one embodiment, search results are automatically reloaded in all directions when, for example, the user comes to the limits of the displayed results.

Further, the module allows a result to be a preview of a hit or a plurality of previews of hits to the base search and the supplemental part-hit search, respectively, on the display. Such a preview shows, for example, only a section of a total hit. For example, the clipping may display or highlight those components that are in accordance with the search. Also, for example, a summary of the hit can be specified. For example, this summary can be retrievable stored in a database. It can also be provided that a summary is created automatically by means of an algorithm which, for this purpose, makes use of the search terms and compiles corresponding relevant contents.

The module can also be provided with a highlighting function. The highlighting feature allows highlighting to display at least the selected part hit on the display. This can be done for example by a kind of magnifying glass, which allows a larger zooming of the hit. The highlighting can also be done by colored background as well as by a combination of different means. For example, an extra window may open on the display in which the partial hit is at least partially displayed. Also, the extra window can open in another display, which is also available to the operator.

An additional embodiment of the module provides that an evaluation module is integrated, by means of which the partial search for at least one to be completed automatically the search term for the partial hit search is evaluated. The evaluation module may preferably have presets. This can be used, for example, to control how the evaluation module should proceed. Thus, for example, not to be considered hits but also terms can be specified. A weighting may also be provided so that a term is then used as a supplementary search term if, for example, it occurs very frequently, is subject to a certain relevance or is classified as suitable by the evaluation module according to another criterion.

For example, the module may have a stored algorithm that implements, for example, one or more of the different requirements described above. For example, an algorithm can be used that works as follows:

In principle, a complete page of a selected search result could be crawled and its text analyzed; but since this is very performance-heavy and the time feedback does not provide good response time behavior for the operator, it is proposed to use only a so-called snippet, a textual short description of a web page, and to analyze it.

Furthermore, there is the possibility that regardless of which texts are used - for example, complete content of the page or a short description thereof - a method is used in which the algorithm proceeds as follows: a list of all contained words is constructed in which all so-called stop words such as meaningless articles, bind words, prepositions and punctuation are filtered out. The rest of the words, mostly nouns, are gestured, that is, words in their plural form become singular, and conjugated and substantiated verbs become verbs. This normalizes all words and makes them comparable in the first place. These are then counted and the number of occurrences within the text is stored next to the normalized version of the word. Then the original search term is ignored, because in this step the search for further, representative, refining terms takes place. For example, the most prevalent word (s) may now be classified as a relevant extension, and a search order up to this point is extended around that word or words.

Further complementary but possibly also possible alternative procedures are designed as follows, whereby these can also be linked to one another or to several others: - A term occurrence frequency is set in relation to a total number of words contained in the text, for example in order to create a relative solution

- a word distribution or accumulation in the text is automatically analyzed and evaluated; if, for example unnaturally 50 times in a row, the same term is in the text, because the author may have wanted to influence the search engine ranking, the weight of that word is automatically corrected downwards, preferably according to an adjustable default, and / or the term is completely ignored

- Lists of terms are given to which certain weights are already assigned

- Differences, especially the greatest possible differences in the content are sought, in order to point out and arrange an orthogonal path of a result or partial result, a really different content direction, which not only a refinement of already thematically obvious content but strong divergent sub-results juxtaposed

In particular, it is possible to obtain a predefinable distinction of individual partial hits. One way to do this is to group the results in a general way. Classification is defined by class and, using one or more algorithms, found results are sorted into these classes. Such a method is used primarily when a manageable number of results is to be expected, for example due to the composition of the data to be searched. If, on the other hand, it is difficult to predict in which area an operator is looking for information, many proposed categories would mostly be useless, empty and unreasonably determined. Therefore, a "clustering" is another variant of a grouping. Clusters of data must know the entire dataset. By an algorithm, for example k-means, a number (k) of clusters is given. Similarity calculations analyze the closeness of the documents to each other and the algorithm sets the limits of similarity between these documents so that, in the end, a certain number of clusters are created, for example exactly 4 clusters that are of different size and / or complexity. In addition, the module or the proposed method influences a data transmission, but in particular also a data utilization, as will be explained in more detail below:

For example, if there is no custom search engine in the background, i. Running in the backend and providing services like the currently used Bing API, every time you select a search result, it will trigger a refining search on that API. Initially, this may not necessarily contribute to the reduction of the communication effort. However, the results of the previous searches as partial paths are also displayed, which makes a possible search for this information unnecessary. About 1/3 of all web searches are repetitive searches. People like to search repeatedly for things they've been looking for on the Internet in the recent past, because they rely on the search engines to retrieve this information at any time. Therefore, hardly stored, printed or gebookmarkt. By using these already created subpaths, the effort is reduced. A further improvement results when using a separate search engine, which runs in the background. Preferably, this is used when a weighting is set. By using the own search engine is strongly worked on the once created index to these data and data queries to search engine interfaces, for example in the form of APIs, as they are usable fall away. This further reduces the data transfer load.

Furthermore, in the context of the invention, the following additional methods implemented in one or more algorithms are proposed, which are preferably implemented in the module:

1. In addition, as well as alternative to the proposed automated termer extension, an implicit search can be triggered by other means:

For example, other embodiments provide for "selecting candidates for implicit search queries / refinements" based on statistical values or comparing them with candidates from the original one to obtain as different types of results as possible Differences in considered groupings, for example, with different weights, some terms may be more important than others. These weights can also adapt themselves according to an embodiment of the user. For example, a Discrete Labeling can be found under Labeling Clusters SOM according to Korinna Bade. It is preferred that a separate search engine runs in the backend, which can also process the weights.

2. Implicit Saving of Search Results:

An operator does not have to actively save search results or bookmark them in English. System-relevant results are automatically set more persistently in time. For example, this can be done by a corresponding visualization: hiding the other results, highlighting the results considered to be interesting by enlargement, for example, a possible parallel reduction of the other results, or Ausgrauen the other results. Criteria for this approach may be varied, for example, measurements of the interaction steps that one has spent on a result, such as zooming in, opening, browsing within the result, are possible. Also, a measurement of the time an operator has spent on a particular result is a frequency of appearance of that one URL as a search result across various searches, and / or potentially important snippet of different searches, or an analysis of the score of that page, e.g. of the PageRank, usable as a relevant measurement. For this purpose, the module can have one or more sensors which are able to detect such and then provide an evaluation device, for example a CPU. This software implementation can also take place in a CPU of the module, so that a sensor system is represented by data evaluation in the module.

An embodiment provides a kind of implicit bookmarking. It leans on the proposed semi-automatic searches described above, so takes the user work, but only acts as an offer to the user that he can or may not perceive. The idea, which is independent and independently weiterverfolg bar, provides a kind of second filter level for "bookmarks" - which is preferably integrated into a search engine interface -but no separate bookmark mode or the like. This eliminates the need for mode changes that would require the user's cognitive abilities.

3. Highlighting join points: A usable, interactive system is characterized according to DIN EN ISO 9241-110 by a high degree of self-describing ability. In order to make the concept of implicit searches more transparent to the user of this application and to better describe the mode of operation, it is proposed to design the method as follows: the search results, the starting points of a refining, implicit search are marked separately by triggering such. For example, a temporary, context-sensitive overlay of these links or nodes is suggested by a kind of "live tile", similar to a Metro-Ul design of Microsoft Windows Mobile 7 or Windows 8, for example with the information "Implicit search results for [Term]".

4. Grouping, e.g. Classification or clustering of the initial results of the explicit search:

For example, an operator searches for "network" and gets the results grouped into different "network" types, such as social networks, technical networks, and others. This serves as a prefilter and allows the operator to specify the domain where it is looking for more information. Here again different visual representations are applicable, for example sets, a substitute for several deposited documents of the same kind, staggered display of first representatives, and others.

5. Fold / Expand Paths:

In order not to waste the screen space, intermediate results, which are considered by the user to be non-relevant, can be presented in a more compressed way. For example, this succeeds by an overlaying arranged arrangement, which, however, can be expanded again at the request of the user.

* 6th Hide over time:

Offered search results, especially those of an implicit search, fade over the usage time and disappear altogether, if they have not been previously marked by the user. This minimizes the so-called "visual clutter" and the screen space will continue to be meaningfully usable and not overloaded, which is comparable to the following metaphor: footprints in the sand by the sea, the tide or the wind let them Time will disappear if the path has not been followed a second time or you are no longer stuck there.

7. Showing relationships among the results:

Another embodiment provides that results are highlighted, which appear in several sub-paths and thus different search contexts. This can be done, for example, by means of a caching algorithm.

7.1 Supporting a transversal axis, according to Laurence Noel in Noel, L. "From semantic web data to inform-action: a means to an end". in proc. of Semantic Web User Interaction at CHI 2008, Exploring HCl Challenges, pp. 1-7. 2008 .: i) It is proposed that a display of domain jumps is generated, which leads the user through a for example automatic analysis of the links of results up to hierarchy level 2 or 3 and an offering of these deeply linked pages is generated by a facilitate easier domain switching. ii) After the xth extension of the initial search term, only the last x extensions for the automatic completion of implicit searches are used, x is configurable by the user. This means, for example, that a user enters "Terml" explicitly in a search mask, successively clicks on interesting results, and thus implicitly searches for extensions.At some point, the query consists, for example, of "Terml Term2 Term3 Term4". Suppose x = 4. Then, after re-refining, "Term2 Term3 Term4 Term5" would be searched to preserve the maximum number of four search refinements, thereby leaving the user, depending on the choice of x, faster or slower in the domain of his or her original search need in the direction of his or her actual search interest.

8. Minimize breach of expectancy compliance:

For this purpose, it is proposed to generate a parallel listing of a conventional list view with conventional result summaries, in English surrogates, in parallel through the module. Thus, for example, the display can be subjected to an automatic split, a crossfade can be generated, as well as a switchover of the various arrangements by the user to be possible. This allows a possibility to be created, a pleasant transition, for example in the form of a crossfade / Animation in the proposed, implemented by the module search method, also called in a version called "Trailblazer" layout to create .This does not allow the expectations of the users to violate too much., Above all, the possibility is created, a learnability of the novel user interface Thus, a familiar representation of a search result can be switched to the linked arrangement generated by the proposed method and automatically complementary search system.

9. Perspective change:

By this is meant that generally alternative arrangements of the results are supported. These can be predefinable or can also be generated by a random generator, which is implemented in the module, for example.

10. Iconic representation of search paths:

In parallel with the exploration of the search paths, a visual, iconic representative of this search builds up and later represents the entire search context, thus allowing the operator an easier re-entry into past research processes.

11. Integration of user feedback, for example, also with regard to collaborative, distributed searches / social search:

For this purpose, it is proposed that a user can manipulate a rank of individual results by rating, for example by a score, by awarding stars, by a thumb up / down. This allows, for example, to change the weighting during the search. There is also the possibility of being able to change a shift in weightings and thus the automatic supplementary part hit selection in a certain search area by a large number of users.

According to a further aspect of the invention, which can also be continued independently as well as in combination with the embodiments described above and described below, a method for carrying out a computer search is proposed with the following steps:

Input of one or more search words into a research module by a

Operator, - first search by the search module based on the entered search words,

Arranging a result of the search module on a display,

Selection of a partial result from the result by the operator,

Search by the search module with the search term (s) from the input supplemented by automatically determined content from the selected partial result,

- Arranging the results of the supplemented search on the display, with the results of the first and supplementary search being at least partially displayed on the screen.

A further development of the method provides that the results from the first and the supplemental search are displayed in different directions on the display.

An example, see FIG. 1, is intended to explain the proposed sequence in more detail. Here, the following interaction sequence is provided:

A - user enters search word in the search field as known from other systems in text form and presses Enter or the search button

B - initially a result list in the form of thumbnails, arranged horizontally and displayed

C - If the user presses on one of these results with, for example, a left-click of the mouse, a second search result row orthogonal to it opens, for example at the height of C, which internally expands the original search by one relevant word from this initial result and this corresponding one new "result list" displays. This can go on as desired - for example, the results are always placed down or going to the right, so the search space, e.g. the WWW, restricted and the user is directed to the information of interest to him.

The advantages are now:

- you do not have to enter a text for a refinement, a click on a promising search result is enough to load more such similar ones

- you can, but you do not have to follow such a subpath

- You can follow them as far as you want

- by kinking the research paths, the user knows exactly where a refinement was made - so you can go back and choose alternatives, - you do not have to open a new window, tab, or other external tool

- all in one view, which provides a good overview, traceability, less susceptibility to errors and reduces the cognitive burden on the user

- The operator works more effectively, efficiently and with higher satisfaction than other search systems.

Further advantageous embodiments, details and features of the present invention are described in detail in the Master Thesis Kathrin Scheil (2012): Trail-based Interaction for Exploratory Search (Master's thesis), submitted on 10.07.2012 to the Faculty of Computer Science of the Otto von Guericke University Magdeburg forth. These are fully referred to in the context of the invention, in particular with regard to a possible implementation of the proposed method, a possible arrangement of result structures and their linking, explanations regarding usable algorithms and other details regarding the proposed module, implemented therein method steps and also different applications and uses. gene.

In accordance with another aspect, a computer program product for performing a computer search method as described above is proposed on a computer. The computer program product can be in the form of a portable data memory, for example as a DVD, as a CD, as a BlueRay, as a data stick, as an SSD hard disk, as a portable hard disk or in any other form. Also, it can be downloaded from a data store, for example, by remote link and brought to execution.

Further advantageous embodiments and examples are explained in more detail with reference to the following figures. The figures serve to illustrate the invention without, however, limiting it. Rather, one or more features of one or more figures having one or more features of other figures as well as of the above description may be linked together to form further embodiments of the invention. Show it:

1 shows an exemplary first representation of a sequence of a proposed search and its refinement together with display on a display,

2 is a schematic view of a module, 3 shows an exemplary representation of an arrangement of a result and a partial search result,

4 to 11 an exemplary sequence of a search with arrangement of the results in the individual steps on the display.

Fig. 1 shows an exemplary embodiment, the individual steps A to D, as starting from a base search A display of the result B on the display, a selection of a hit by a user, then automatically by the module a partial hit result C in different ways to the result of the basic search is displayed, whereby a combination of both results has been preserved. D shows further possibilities, how the representation of the results to each other can be made possible.

It is therefore an interactive user interface, the English User Interface, which allows an ergonomic semi-adaptive display of search results. It serves the ergonomic, more effective, more efficient and with higher user satisfaction associated use of search engines or information retrieval systems. The known from the prior art disadvantages such as u.a. lack of representation of relationships between search results, the need for explicit search refinement by e.g. Widening the request for additional terms, frequent mode changes, thereby disproportionately high cognitive load on the user, are avoided by the proposed approach and the proposed module. Rather, these facilitate the interactive information search with a computer. In particular, longer-term search tasks, so-called exploratory searches whose destination is still unknown at the beginning of the search, can be supported. Here, for example, the visualization of search results in the form of page previews is used and also the arrangement of the results next to or below each other.

Furthermore, the proposed method as well as the module has numerous additional advantages and properties:

(a) implicit request formulation: instead of requiring the user to extend the original search term, the query is indirectly expanded by representative terms by clicking on a search result that appears attractive to the user; these terms the user did not have to be able to name before; the "extended" Results are preferably arranged orthogonal to the selected search result element, for example, down or right branching off as a new result list;

(b) through this domino game-like organization of the result space, unique paths, in English trails, are built up, which can be stored as pictorial representatives to secure search or search histories, cf. 1, D, and because of their easily recognizable shape can serve the user as a map;

(c) Moving paths, for example to the top left, creates the possibility of carrying out parallel searches. For example, no page replacement, tabs, windows or third-party programs are required; This saves unergonomic context and mode changes. On the other hand, Zoomin / ZoomOut, inframe browsing for exploration and other means may be preferred.

Fig. 2 shows a schematic view of a module 1, which has various individual components which are spatially separated from each other. The module 1 with an interactive representation 2 of a search query on a display 3 for an operator for a computer 4, comprises:

a search module 5, which makes an input of a search available and on the basis of which the search module carries out a basic search,

the display 3, which visually displays a search result 6 for the base search,

a selection option 7, which displays a partial hit of the search result and makes it possible to select the partial match,

a supplemental part-hit search 8, which, when the part hit is selected, independently carries out an additional search by means of the search module, whereby a content of the part hit is automatically included in the additional search, and

an arrangement 9 of a result of the supplementary part-hit search on the display, which differs from an indication of the search result of the search on the display 3.

The selection option is symbolized here, for example, as an arrow controlled by a mouse 10. By contrast, a computer program product is shown here by way of example in the form of a DVD with which a method can be implemented on the computer 4. 3 shows an exemplary representation of an arrangement. Furthermore, various control options emerge from the view, which the operator can use during the search, for example to select a partial hit. All results are displayed as an overlay so that the user has an overview of the result space. A change from the Start screen to display the result is also animated. For example, the title and URL are always displayed. In addition, a text-based mode can also be activated. This is indicated by the result snippets, between the title and the URL. For example, this text will also be displayed if a partial hit has not yet been loaded so that results are not empty.

Preferably, the layout provides the creation of icons that can replace, for example, text-based buttons. A new arrangement over the header area results from the size of the buttons. The icons are slightly larger for a tablet version to accommodate the coarser touch interaction. Such icons are exemplified as a set at the top of FIG. For example, the iconset is based on Entypo45, a set of pictograms released under the Creative Commons License CC BY-SA46. The pictograms can have different functions and thus enable a different operation. For example, it is provided that deactivated keys are grayed out.

For example, all subtotals can be deleted from the result using the key that displays the trash can. This key is preferably highlighted in red. Another feature is the two diagonal arrow symbols, which are preferably blue. These are used to jump from one explicit query to another. In the process, a certain search term is loaded into the search term input and the change highlighting is renewed in order to promote understanding. The next two icons, which are preferably highlighted in green, allow you to switch between gently animated thumbnail and text-based displays. This is an example of an alternative to a zoom function on the mouse wheel, which may have the scrolling of the result space imaged on the wheel. The mouse symbol, which is preferably highlighted in green, makes it possible, for example, to change this assignment. The last button, preferably highlighted in green, shows an information dialog describing the possible interactions with a result.

The additional search terms after the output of an implicit query can be separated according to a further development by means of the previous query, so that the Gives users a better overview of automated search term extension. This means that in the partial search automatically supplemented search word or search words is or will be displayed to the operator. Another embodiment provides that, for example, in a preselection as well as in between typical settings such

Wildcards, Boolean links and others are included in the automated supplement for the partial research.

In the following, the general procedure of a standard user interaction is illustrated by means of dialog examples in order to get an impression of the realized layout solution without access to the program. The following explanations are based on the corresponding preceding figure, respectively.

The startup screen of FIG. 4 is a simple layout for entering a query. The example given was "financial crisis europe".

After the query is issued, the first results - the number of results depends on the screen width - are displayed. Some navigation and function keys are located on the upper right side, cf. Fig. 5.

When dialing a partial hit is scaled slightly, cf. Figure 6. When a user selects the second result, it will be displayed in preview mode and the result will be highlighted. To open the result in a new window, to enlarge it or to return to the overview, from left to right, there are three more function keys on the upper right side of the preview window. After clicking on the enlargement key, a full-screen mode is activated in which the user can handle the hit in the usual way, cf. Figure 7. When the user decides to return to the overview, the implicit sub-results are added below the selected result. The selected result is displayed on the upper left corner of the result area, cf. Figure 8. When selecting an implicit hit on the vertical axis, another implicit hit can be retrieved, which is seen on the right. After returning from the preview mode, further implicit results are shown on the horizontal axis. If the website preview has not yet been created or loaded, the textual preview is displayed, cf. Fig. 9 and Fig. 10. The user can activate the textual preview to display only the title, a text snippet and the URL for each hit, cf. Figure 11.

Claims

claims

1. Module with an interactive representation of a query to an information retrieval system, via a computer on a display for an operator, comprising:

a search module which makes an input of a search available and on the basis of which the search module carries out a basic search,

the display which visually displays a search result for the basic search,

an arrangement of a result of the supplementary part-hit search on the display, which differs from an arrangement of the search result of the previous search on the display.

2. Module according to claim 1, characterized in that an evaluation module is present, by means of which a partial search is automatically evaluated for the partial search for at least one search term to be supplemented for the partial hit search, wherein the evaluation module has presettings, via which it is controlled, in which way the evaluation module proceeds, the default settings include:

- hits to be considered,

- Predefinable word terms and / or

a weighting so that a word term is used as a supplementary search term, preferably a weighting, so that individual words or terms which could be used as supplementary search term (s) have a certain weight for influencing the further search history can be assigned ..

3. Module according to claim 1 or 2, characterized in that the arrangement of several results of the supplementary partial search on the display is orthogonal to the search result of the base search.

4. Module according to claim 1, 2 or 3, characterized in that the arrangement of different results of consecutive searches is structured tree-like.

5. Module according to one of the preceding claims, characterized in that the arrangement of at least one result on the display by the operator is displaceable.

6. Module according to one of the preceding claims, characterized in that a result is a preview of a hit or a plurality of previews of hits to the base search and the supplemental partial hit search on the display.

7. Module according to one of the preceding claims, characterized in that a highlighting function for displaying at least the selected part hit is present on the display.

8. Module according to one of the preceding claims, characterized in that an evaluation module is integrated, by means of which the partial search is evaluated for at least one supplementary search term for the partial hit search, preferably that an evaluation module is integrated, by means of which automated the partial hit search for previous search and at least one search term to be supplemented automatically for the partial hit search.

9. A method for performing a computer search, preferably with a module according to one of the preceding claims, comprising the following steps:

Input of one or more search words into a search module by an operator,

- first search by the search module based on the entered search words,

Arranging a result of the search module on a display,

Selection of a partial result from the result by the operator,

automated research by the search module with the search term (s) added to the input, supplemented by automatically determined content from the selected partial result,

Arrangement of the results from the supplemented search on the display, wherein the results of the first and supplemental search are each at least partially displayed on the display and shown differently from each other, but linked together.

10. The method according to claim 8, characterized in that the results of first and supplemented research are displayed in different orientation on the display.

A computer program product for carrying out a computer search method according to the features of one of claims 8 or 9 on a computer.