Product IP Search: Constructing a “Google for Engineers” Framework

During the past five years, search engine technology has become a significant component in the enterprise IT arsenal applied to a wide variety of applications and use cases. Search engines are now commonplace in applications such as customer-facing Web portals, document management, data mining, and many business oriented applications such as ERP and CRM.  But can search engines be usefully applied to technical and engineering oriented use cases?  This article outlines the potential benefits and unique challenges associated with engineering oriented, Product IP search and discusses several classes of potential solution vendors in this area.

Potential Benefits
There are three major potential benefits of deploying Product IP Search including:

• Better decision making leading to increased product innovation and/or lower product cost
• Increased productivity
• Lower IT costs

Better Decision Making
When developing products, millions of decisions are made daily involving a whole host of parameters which affect a product’s features, performance, quality, maintainability, cost, etc. In many cases, these decisions are made with a less-than-complete set of facts or knowledge, primarily because decision makers do not have full and ready access to the universe of all relevant information elements. A simple engineering example would be the decision factors surrounding the selection of a particular fastener component. Engineers typically have a good idea on the engineering considerations involved with such a decision, but often times do not have access to the cost, reliability, inventory, and maintainability factors because this data resides in systems which are not easily accessible. So the engineer may decide on using a fastener which will fulfill the engineering requirements but is sub-optimal from a cost standpoint.  Another problem that often arises centers on the latency or lag time of information moving from one point to another. A common example is the latency which occurs from customer warranty claims or product recalls to the point in time where corrective measures are actually designed into the product.  If an engineer had complete visibility early on to potential product issues in the field, the latency between product reliability issues and corrective action could be dramatically reduced. The key therefore to making better decisions is to provide access to a richer, wider universe of relevant information and to reduce the latency of getting that information into the hands of decision makers. Search engines are an ideal mechanism for accomplishing this goal.

Increased Productivity
The engineering technology vendors have done a wonderful job in creating systems and tools for authoring, validating, and manufacturing products leading to sizeable engineering productivity gains. However, when engineering users need to access data and information residing outside of their CAD/CAM/PLM systems, productivity often drops off dramatically due to the hodgepodge collection of systems and inefficient mechanisms used for accessing this data. Using a search engine with a common, easily accessible Web type interface can dramatically improve the ease of finding and retrieving data which reside in numerous external files and data repositories.

Lower IT Costs
Many companies, particularly the larger Fortune 500 type firms, have developed custom systems or mechanisms which allow engineering type users to access a broader array of Product IP content. These in-house systems usually consist of either point-to-point interfaces between major enterprise systems (e. g. PLM, ERP) and smaller legacy systems, along with data migration activities and perhaps some type of data warehouse. This type of data aggregation is very expensive and  IT intensive from a support standpoint. Search engines can potential lower IT costs rather significantly here by providing a much simpler front-end access mechanism while virtually eliminating the need for any data migration.

Key Product IP Search Requirements and Challenges
Applying search engine technologies to find, retrieve and navigate Product IP introduces several unique challenges when compared to generic text searching including:

• Complex data types and relationships
• Product aware indexing taxonomies
• Dynamic Data classification, sorting and navigation
• Engineering system connectors

Complex Data Types and Relationships
Digital Product IP consists of a wide range of engineering-oriented content containing both simple and complex data types. Examples of digital Product IP include:

• CAD/CAM/CAE graphics files
• PDM/PLM data
• Flat image file data such as drawings
• Technical documents in areas such as testing and quality
• Specifications and BOM data
• Part and component data bases
• Numerous specialty databases in areas such as maintenance

Due to their document-centric, keyword heritage, few commercial enterprise search engines are equipped to deal effectively with some of the more complex engineering data types such as indexing CAD data. In addition, the relationships of data in the Product IP world are equally if not more important than the underlying data itself. Knowing for example how potential changes to an underlying component affect all of the related products and assemblies using this component is vital to engineering decision making. So it is absolutely essential that the Product IP search engine has the ability to not only index information contained in a particular document or data set, but to capture and understand the relationships surrounding it.

Product Aware Indexing Taxonomies
Product IP contains a unique taxonomy and set of nomenclature that when applied correctly, can help to create a more useful index by normalizing equivalent key words and phrases (e. g. version and revision, part and component) and help to better characterize the content of specific data and documents. In addition, specialized geometric shape and proximity search mechanisms are now emerging which can infer and make explicit, very useful data attributes for searches involving geometric or product characteristics.  Having a rich product aware taxonomy is a critical component for robust Product IP searching.

Dynamic Data Classification, Sorting and Navigation
Many Product IP use cases center on a progressive discovery style of searching. This involves first retrieving a broad and potentially large data set and subsequently sorting and narrowing the search result into a more targeted, navigable data set. For example, a user may wish to find information on mechanical fasteners. The first cut search result may result in a million returned hits. Ideally, the user would want to classify and sort this data by key attributes (e. g. size, material, supplier, inventory levels) and then have the ability to navigate and narrow the search to fit relevant criteria. The ability to place and sort data into various attribute buckets on the fly is called dynamic classification. A number of commercial search engines now include this ability to dynamically sort the contents of search results without requiring any pre-programming or specialized data tagging. This allows users to generically understand the underlying characteristics of returned data sets at a high level which in turn helps to refine, navigate to, and access the exact data they require.

Engineering Systems Connectors
One of the more difficult challenges in creating a robust Product IP index is that an important percentage of the raw data to be indexed resides in numerous and incompatible, specialized engineering systems. While index connectors for common data and document types are widely available (e. g. MS Office Documents, generic databases), connectors which can reliably and efficiently index the myriad number of engineering systems are rare and difficult to build.

Potential Solution Vendors
There are three identifiable classes of vendors which are marketing solutions or portions of solutions in this area:

• Enterprise Search Vendors
• PLM Vendors
• Specialty Search Vendors

Enterprise Search Vendors
Enterprise search vendors include companies such as Google, Autonomy and Endeca each of which are beginning to market their generic search engine solutions into the engineering and manufacturing vertical market. Currently, these vendors offer very limited Product Aware indexing capabilities and little if any specialized engineering connectivity. However, there are signs that these vendors are starting to develop richer Product IP capabilities through targeted alliances and internal R&D efforts.

PLM Vendors
All of the major PLM Vendors such as Dassault, PTC and UGS are actively developing and/or partnering with search engine companies to deliver robust Product IP search solutions. The solutions delivered so far primarily focus on search within the boundaries of their own respective PDM/PLM systems and data that has been linked into these systems. During the next two years, these vendors are likely to deliver solutions which more closely conform to the Product IP ideal.

Specialty Vendors
The most advanced vendor with the broadest set of Product IP search capabilities is Centric Software which has built a solution (Insight) focused exclusively in this area. There are also a number of smaller specialty vendors which provide pieces of the Product IP solution in areas such as engineering system connectivity. These vendors include companies such as ProStep and Right Hemisphere.

Summary
The benefits of applying search-based technologies in Product IP domains are significant and potentially large. While there are several significant technical challenges and hurdles to clear for realizing these benefits, none of these are showstoppers, and are well within the current technology trajectories of vendors pursing this opportunity.